United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT
Argued April 17, 2000 Decided September 12, 2000
No. 99-5110
George E. Cone, Jr.,
Appellee
v.
Louis Caldera, Secretary of the Army,
Appellant
Appeal from the United States District Court
for the District of Columbia
(No. 97cv00168)
Fred E. Haynes, Assistant U.S. Attorney, argued the cause
for appellant. With him on the brief were Wilma A. Lewis,
U.S. Attorney, and Michael J. Ryan and R. Craig Lawrence,
Assistant U.S. Attorneys.
John A. Wickham argued the cause and filed the brief for
appellee.
Before: Williams, Randolph, and Garland, Circuit Judges.
Opinion for the Court filed by Circuit Judge Garland.
Garland, Circuit Judge: George E. Cone, Jr. filed suit in
United States District Court challenging the Army's refusal
to amend his Officer Evaluation Report. The court concluded
that the Army's refusal was arbitrary and capricious, and
directed the Army to improve Cone's rating. We reverse.
I
Cone served as a captain in the United States Army and
commanded an infantry company during the Persian Gulf
War.1 After the war, his supervisors completed an Officer
Evaluation Report (OER) assessing his performance from
June 14, 1990 to June 13, 1991. An OER is used to evaluate
an officer's performance and career potential. See Army
Regulation 623-105, at p 1-6(a) [hereinafter AR 623-105]. At
least two of the officer's superiors prepare the report--a
"rater" who directly supervises the officer and a "senior
rater" higher in the chain of command. Id. p 3-1. Cone's
rater was his battalion commander, Lt. Colonel Stephen S.
Smith. His senior rater was his brigade commander, Colonel
James C. Riley.
The OER form contains blanks that require the rater and
senior rater to provide both numerical and narrative assess-
ments. Part VII of the form is reserved for the senior rater,
who evaluates the officer "by comparing the rated officer's
potential with all other officers of the same grade." Id.
p 4-16(b). In Part VII(a), the senior rater is to check one of
ten blocks that rank the officer's potential against that of all
other officers of that grade. According to Army regulations,
the evaluation is "based on the premise that in a representa-
tive sample of 100 officers of the same grade or grade
grouping (Army-wide), the relative potential of such a sample
will approximate a bell-shaped normal distribution pattern."
Id.2 This means, the regulations continue, "that in a repre-
__________
1 Cone was promoted to major in October of 1996.
2 In a normal distribution, most of the data points are "clus-
tered near the mean, and the density or relative frequency of the
sentative sample of 100 officers of the same grade or grade
grouping (Army-wide) only one officer can reasonably be
expected to be placed in the top block." Id. p 4-16(c).
Slightly more officers may be ranked in the second block,
which represents the top 2-3% of Army officers of that grade,
and so on. In a representative sample of 100, it is expected
that the majority of officers will be rated in the two middle
blocks. See J.A. at 147 (illustrative OER form). The senior
rater must also write narrative comments addressing the
officer's performance in Part VII(b) of the OER. See AR
623-105, at p 4-16(d)(2), (3).3
Once completed, the Army compares the senior rater's
assessment of the officer to the senior rater's rating history
for all officers of the same grade--known as the senior rater's
"profile." Id. pp 2-5, 4-16(d)(5)(a). "The purpose of the
profile is to place the rated officer's OER in perspective by
revealing the senior rater's general rating tendency." Id.
p 4-16(d)(5)(a). By comparing a specific OER to the profile,
the Army can discern whether a particular officer performed
above, at, or below the "center of mass"--i.e., the median
ranking--of the officers ranked by the same senior rater.
In completing Cone's OER, senior rater Riley checked the
second block from the top. Of eleven captains rated by Riley
during this period, he ranked seven in the top block, three
(including Cone) in the second block, and one in the third
block. See J.A. at 147. According to this profile--which was
included in Cone's OER--Riley's center-of-mass ("COM")
__________
numbers decreases with increasing distance from the mean."
David W. Barnes & John M. Conley, Statistical Evidence in
Litigation 140 (1986). "If the frequencies with which the numbers
in the normal population appear as the result of random experi-
ments are plotted on a graph, the figure resembles a bell-shaped
curve." Id.
3 The Army has recently altered its regulations and OER form,
and has dispensed with the ten-block rating system. See Army
Regulation 623-105 (Apr. 1, 1998); DA Form 67-9 (Apr. 1, 1998).
The citations in this opinion refer to the regulations and forms that
were in effect at the time of Cone's evaluation.
was the top block. Cone's rating was therefore below center-
of-mass for the group even though he was rated in the
second-highest block, which corresponds to the top 2-3% of
captains Army-wide.
Cone appealed his OER to the Officer Special Review
Board (OSRB), which is charged with reviewing requests to
correct erroneous military records. See AR 623-105, at
p 9-2(i). He contended that "[t]he senior rater blocking ...
does not accurately portray my potential and may penalize
me when viewed by future boards." J.A. at 144. As support
for this assertion, he supplied affidavits from Riley and Smith
indicating that Riley "did not intend for [Cone's] OER to
reflect below center of mass performance," and acknowl-
edging that Riley had failed "to tightly manage" his senior
rater profile. Id. at 150.
The Special Review Board denied Cone's request. See
OSRB Decision at 4 (July 2, 1993) (J.A. at 143). The Board
rejected the affidavit from Riley, noting that Army regula-
tions preclude "statements from rating officials that they did
not intend to evaluate as they did from serving as the basis
for favorable action on appeals." Id.; see AR 623-105, at
p 5-32(b)(2), App. p N-2(b)(3). The Special Review Board
also noted that Riley had submitted "virtually identical"
memoranda in support of appeals by five other captains he
had similarly rated below center-of-mass. 1993 OSRB Deci-
sion at 2 (J.A. at 141).4 Although the Board recognized that
there "was a persistent failure by the [senior rater] to man-
age his profile for [captains] in such a way as to preclude any
unintended below-COM ratings," it concluded that Cone had
failed to adduce "clear and convincing evidence" (as required
by the regulations) to rebut the "presumption of regularity"
in the OER. Id. at 3-4 (J.A. at 142-43) (citing AR 623-105,
at p 5-32).
__________
4 The Special Review Board further found that of nine below
center-of-mass ratings made by Riley and processed by the Army
between August 1990 and February 1992, he supported or offered
to support appeals by at least seven of the subject officers. See
1993 OSRB Decision at 3 (J.A. at 142).
Relying on new statements from his rater and senior rater,
Cone once again appealed to the Special Review Board and
the Board once again denied Cone's request. See OSRB
Decision (July 29, 1994) (J.A. at 158-60). In particular, the
Board rejected Cone's claim that the comments Riley wrote
in the narrative portion of the OER supported a center-of-
mass rating. Those words, the Special Review Board said,
"are not comparable to" what Riley wrote regarding "other
officers [whom] he did in fact place in the top block COM."
Id. Specifically, the Board noted that Riley characterized
Cone as having "solid potential," while he described those he
placed in the top block as having "unlimited" or "outstanding
potential." Id. Similarly, while Riley's recommendation re-
garding Cone was "promote to major and consider for [Com-
mand and General Staff College] when eligible," for the
others it was "promote to major and select for [Command and
General Staff College] ASAP." Id. (emphasis added).
Cone next appealed to the Army Board for Correction of
Military Records. See 10 U.S.C. s 1552(a)(1). Based on the
same evidence, Cone urged amendment of his OER to reflect
a center-of-mass rating. The Correction Board rejected his
appeal, concluding that, despite Riley's inability "to establish
a ... profile for captains that approximated a bell-shaped
normal distribution," Cone's rating accurately captured Ri-
ley's assessment of Cone's performance compared to that of
other officers of the same grade. Correction Board Decision
at 4 (Apr. 26, 1995) (J.A. at 128). The Board noted that its
conclusion was also "supported by the type of comments used
by the [senior rater] to distinguish between a top block officer
and a two block officer." Id.
After losing before the Correction Board, Cone filed suit in
the United States District Court for the District of Columbia,
alleging that the Board's denial of his request violated the
Administrative Procedure Act, 5 U.S.C. s 706(2)(A), and ask-
ing the court to correct his records.5 Specifically, Cone
__________
5 Cone also challenged his nonselection to the Command and
General Staff College, a challenge the district court found non-
reviewable. See Cone v. Caldera, 46 F. Supp. 2d 3, 7-8 (D.D.C.
argued that Army regulations require a senior rater to rank
officers along a bell-shaped curve, and that Riley's failure to
maintain a profile hewn to that curve constituted arbitrary
and capricious action. Although the district court found that
the regulations "do not mandate that each senior rater main-
tain a bell-shaped-curve distribution," Cone v. Caldera, 46
F. Supp. 2d 3, 6 (D.D.C. 1999) (internal quotation omitted), it
nevertheless held the Army's ranking arbitrary and capri-
cious, see id. at 7. After undertaking a statistical analysis,
the court determined that "[t]he senior rater's ranking of
seven captains in the top block is more than twenty standard
deviations above the expected value," and that "the center of
mass of an eleven-officer profile" could not "reasonably be
located in the top block." Id. at 6-7. This, the court said,
indicated that the Army had acted arbitrarily and capriciously
in declining to amend Cone's OER "to reflect a center of
mass rating." Id. at 7. The court ordered the Army to make
the necessary amendment, see id., and the Army appealed.
II
The Secretary of a military department, acting through a
civilian board, "may correct any military record of the Secre-
tary's department when the Secretary considers it necessary
to correct an error or remove an injustice." 10 U.S.C.
s 1552(a). An officer's OERs are presumed to be "adminis-
tratively correct" and to "[r]epresent the considered opinions
and objective judgment of the rating officials at the time of
preparation." AR 623-105, at p 5-32. An officer seeking a
correction must prove "clearly and convincingly" that the
"presumption of regularity" in the preparation of administra-
tive records should not apply, and that "[a]ction is warranted
to correct a material error, inaccuracy, or injustice." Id.
p 9-7(a) (citing id. p 5-32); see Frizelle v. Slater, 111 F.3d
172, 177 (D.C. Cir. 1997) (noting "strong but rebuttable
__________
1999). Cone has not pressed this claim on appeal, apparently
because it became moot due to his subsequent selection to the
College. See Cone Br. at 4.
presumption that administrators of the military, like other
public officers, discharge their duties correctly, lawfully, and
in good faith") (internal quotations omitted).
Although we have jurisdiction to review the decisions of the
Correction Board,6 we do so under an "unusually deferential
application of the 'arbitrary or capricious' standard" of the
Administrative Procedure Act. Kreis v. Secretary of the Air
Force, 866 F.2d 1508, 1514 (D.C. Cir. 1989); see Kidwell v.
Department of the Army, 56 F.3d 279, 286 (D.C. Cir. 1995).7
This deferential standard is calculated to ensure that the
courts do not become a forum for appeals by every soldier
dissatisfied with his or her ratings, a result that would
destabilize military command and take the judiciary far afield
of its area of competence. See Orloff v. Willoughby, 345 U.S.
83, 94 (1953) ("Orderly government requires that the judicia-
ry be as scrupulous not to interfere with legitimate Army
matters as the Army must be scrupulous not to intervene in
judicial matters."). We review de novo the district court's
ruling, on cross-motions for summary judgment, that the
Correction Board acted arbitrarily and capriciously in this
case. See Frizelle, 111 F.3d at 176; Kidwell, 56 F.3d at 286.
Cone contends that Riley's failure to adhere to a bell-curve
distribution, and the Army's refusal to correct that failure,
constituted arbitrary and capricious agency action in violation
of Army regulations. The regulations, however, do not re-
quire adherence to a bell-shaped curve; rather, they state a
"premise" upon which senior raters are to base their evalua-
tions:
__________
6 See Kreis v. Secretary of the Air Force, 866 F.2d 1508, 1512
(D.C. Cir. 1989) (citing Chappell v. Wallace, 462 U.S. 296 (1983));
see also Frizelle, 111 F.3d at 176; Dickson v. Secretary of Defense,
68 F.3d 1396, 1401-02 & n.6 (D.C. Cir. 1995); Kidwell v. Depart-
ment of the Army, 56 F.3d 279, 283-86 (D.C. Cir. 1995).
7 Cone's appellate brief asserts that his incorrect OER also
violated his rights under the Privacy Act, 5 U.S.C. s 552a(e)(5).
Neither Cone's complaint nor the district court's opinion addresses
such a claim, and we therefore do not consider it here.
The senior rater's evaluation is made by comparing the
rated officer's potential with all other officers of the same
grade, or grade groupings.... His or her evaluation is
based on the premise that in a representative sample of
100 officers of the same grade or grade grouping (Army-
wide), the relative potential of such a sample will approx-
imate a bell-shaped normal distribution pattern.
AR 623-105, at p 4-16(b). As the district court itself said,
this regulation does "not mandate that each senior rater
maintain a bell-shaped-curve distribution" in each rating peri-
od. Cone, 46 F. Supp. 2d at 6 (internal quotation omitted).8
Nonetheless, the district court concluded that Riley's rank-
ings of his eleven officers departed so radically from what
statistically would have been expected as to prove that Riley
"violated p 4-16 in rating at least some of the other captains
against whom Cone was compared." Id. To reach this
conclusion, the district court applied a test of statistical
significance to the distribution of Riley's rankings of the
eleven captains. See id. at 6 & n.1.
Even without a statistical analysis, it is clear that Riley was
lax in following the rating guidelines. The Special Review
Board itself determined that, over a series of rating periods,
Riley exhibited "a persistent failure ... to manage his profile
for [captains] in such a way as to preclude any unintended
below-COM ratings." 1993 OSRB Decision at 3 (J.A. at 142).
But Riley's inability to adhere to a bell-curve distribution is
not sufficient to render arbitrary the Army's refusal to cor-
rect Cone's record. The Army anticipates, and compensates
for, the fallibility of individual raters by requiring that each
rater's personal profile (indicating the number of officers he
or she has placed in each box during the rating period) be
included in the OER of each officer he or she reviews. As
__________
8 See also Cone, 46 F. Supp. 2d at 6 ("The Army correctly notes
that ... 'since there are thousands of captains in the Army, it is
statistically possible that a particular senior rater may have a
number of captains that he or she is senior rating that merit
placement in the top block."') (quoting Def.'s Resp. to Pl.'s State-
ment of Facts p 41).
the regulations state, "[t]he purpose of the profile is to place
the rated officer's OER in perspective by revealing the senior
rater's general rating tendency." AR 623-105, at
p 4-16(d)(5)(a). Put more bluntly, "[t]he profile is meant to
be an indicator of the senior rater's tendency to inflate or
deflate ratings." Id. p 9-7(f). Hence, anyone examining the
OER to determine the true ability of the officer can take into
account whether his ratings have been rendered by an "easy"
or "hard" grader. This is a reasonable response to the
problem, and one which makes the Correction Board's refusal
to amend Cone's OER reasonable as well.
Moreover, we would be hard pressed to find a reversible
"error" in Cone's record even without the clarification provid-
ed by the senior rater's profile. Under the regulations, the
senior rater's obligation is to compare "the rated officer's
potential with [that of] all other officers of the same grade"
Army-wide. Id. p 4-16(b). The rating box in which Cone
was placed represents the top 2-3% of all Army officers of
the same rank. There is no evidence in the record that
Cone's true abilities place him in the only higher category--a
category reserved for the Army's top 1%. See id. p 4-16(c).
Indeed, the principal flaw to which Cone points--Riley's
inability to adhere to a bell curve--indicates only that Riley
tended to inflate his officers' ratings above the curve: he
consistently placed most of his officers in the top few boxes
on the rating chart, despite the instruction that those boxes
be reserved for the top few percent of all officers in the
Army. See, e.g., 1993 OSRB Decision at 3 (J.A. at 142)
(noting profile periods in which Riley gave top-block ratings
to 22 of 61, 12 of 19, 6 of 11, and 26 of 32 captains). Although
this could suggest that Cone's ranking should not have been
as high as the one Riley gave him, it certainly does not
establish that he should have been ranked even higher.9
__________
9 Cone notes that Riley wrote on the OER: "This evaluation
does not reflect a downturn in performance, rather I have restarted
my profile." J.A. at 147. Army regulations permit a rater to
restart his profile when necessary to convey his intended evaluation
to selection boards and personnel managers. See AR 623-105, at
p 3-12.1. Although Riley's comment means that Cone's second-
In support of the contention that he was misrated, Cone
offers a post-evaluation affidavit in which Riley states that he
"did not intend for [Cone's] OER to reflect below center of
mass performance." J.A. at 150. But the Correction Board
reasonably refused to credit the affidavit on the basis of
Army regulations providing that "[s]tatements from rating
officials that they did not intend to rate [an officer] as they
did" will "not be used to alter or withdraw a report." AR
623-105, at p 5-32(b)(2); see id. App. p N-2(b)(3). The ratio-
nale for these regulations is an understanding that raters
may attempt to retract otherwise accurate assessments when
requested to do so by their disappointed officers. Such post-
hoc statements by rating officials, the regulations explain,
"often reflect retrospective thinking, or second thoughts,
prompted by an appellant's non-selection or other unfavorable
personnel action claimed to be the sole result of the contested
report." Id. The Correction Board's refusal to credit Riley's
affidavit is particularly reasonable in this case, given that
Riley had submitted "virtually identical" memoranda in sup-
port of the appeals of many other captains who likewise
challenged the below center-of-mass ratings he gave them
during the same time period. 1993 OSRB Decision at 2 (J.A.
at 141). The Board reasonably regarded these memoranda
as reflections of Riley's sympathy for the pleas of his subor-
dinates, rather than as accurate statements of his original
intent.
Cone also asks us to consider affidavits, authored by senior
Army officers, that contend that the adjectives Riley used to
describe him in the narrative section of the OER indicate a
center-of-mass rating. See J.A. at 96-99. Both affidavits
were written in November of 1998--after the district court
complaint was filed and long after the close of the administra-
tive proceedings. Because the affidavits were not part of the
administrative record, we may not consider them on appeal.
See Environmental Defense Fund, Inc. v. Costle, 657 F.2d
__________
block rating should not be regarded as a downturn in Cone's own
performance, it says nothing about where he ranks in comparison
with his peers.
275, 284-86 (D.C. Cir. 1981); see also Camp v. Pitts, 411 U.S.
138, 142 (1973). In any event, the affidavits are neither
sufficiently specific, nor sufficiently authoritative, to affect the
result here.10
In conclusion, we observe that the district court's remedial
order reveals just what a quagmire we would enter were we
to uphold Cone's challenge to his OER. The court directed
the Army to amend Part VII(a) of Cone's report "to reflect a
center of mass rating." Cone, 46 F. Supp. 2d at 7. But how
should it be amended? Cone interprets the court's mandate
as requiring that he be given a top-block rating, see Cone Br.
at 34-35, as do we.11 Yet, as already noted, there is no
evidence that Cone qualifies as one of the Army's top 1%.
Moreover, even if Riley were guilty of rampant grade-
inflation, there is no evidence that the seven officers to whom
he gave top-block rankings still did not merit higher rankings
than Cone. Thus, if Cone were moved to the top block to
achieve a center-of-mass result, would not those seven officers
have grounds for appeal, complaining that their own compara-
tive positions had been unfairly deflated? Indeed, would not
the same be true for all captains, Army-wide, whose senior
__________
10 By the same token, we decline to give any weight to the
Special Review Board's conclusion that the words Riley used in
Cone's OER were "not comparable" to the words he used to
describe other officers whom "he did in fact place in the top block
COM." 1994 OSRB Decision (J.A. at 159). The Army successfully
argued below that the court should deny Cone discovery of the
OERs of the other officers upon which this comparison was based,
even in redacted form, on the ground that they were not part of the
administrative record before the Correction Board. See Cone v.
Caldera, Civ. No. 97-168, slip op. at 2-3 (D.D.C. Aug. 21, 1998) (J.A.
at 33-34). But if that is true, then the comparisons cannot provide
substantial evidence to support the Board's decision. On the other
hand, if the comparisons were before the Board, by keeping them
out of the appellate record the Army has deprived this court of any
basis for concluding that the Board's characterizations are valid.
11 The other alternative--lowering the ratings of some officers
now in the top block in order to make the second block the center-
of-mass--would only further complicate the problem discussed in
the following text.
raters had placed them in the top block? The knowledge that
"correcting" Cone's grades would ultimately require us to
reassess the relative rankings of his entire cohort--a task for
which a court is spectacularly unsuited--only further con-
firms the wisdom of deferring to the reasonable judgment of
the Correction Board. See Kreis, 866 F.2d at 1514 ("[T]he
alteration of a record may correct one injustice only to
commit another, or perhaps only to incur some other equally
significant institutional cost. All such balancing of consider-
ations is to be done by the Secretary, free of judicial second-
guessing."); see also Sargisson v. United States, 913 F.2d
918, 922 (Fed. Cir. 1990) ("A court lacks the special expertise
needed to review reserve officers' records and rank them on
the basis of relative merit.").
III
At bottom, Cone argues not that he was graded too low in
comparison to all officers of his rank, but that he did not get
the full benefit of his superior's inflated grading curve. Be-
cause we conclude that this does not constitute cause for
requiring the Correction Board to amend Cone's evaluation
report, we reverse the decision of the district court.