Teacher Ratings of ADHD Symptoms in Ethnic Minority Students:

Bias or Behavioral Difference?

Shelley J. Hosterman and

George J. DuPaul

Lehigh University

Asha K. Jitendra

University of Minnesota

Disproportionate placement of African American and Hispanic students into disability
and special education categories may result from true behavioral and cognitive differ-
ences, bias in assessment and referral, or some combination of the two. Studies of
commonly used ADHD rating scales suggest teacher bias may contribute to placement
discrepancies. This investigation compared teacher ratings of ADHD symptoms on the
Conner’s Teacher Rating Scale—Revised Long Version (CTRS-R:L; Conners, 1997)
and the ADHD-IV: School Version (DuPaul, Power, Anastopoulous, & Reid, 1998),
with objective classroom observations from the Behavioral Observation of Students in
Schools code (BOSS; Shapiro, 2003). Participants were first through fourth grade
students (N

⫽ 172; 120 male) classified as Caucasian (n ⫽ 112) or ethnic minority (17

African American, 38 Hispanic, 5 African American and Hispanic). Contrary to
hypothesis, results showed teacher ratings of ethnic minority students were more
consistent with direct observation data than were ratings of Caucasian students.
Findings suggest teacher ratings of ethnic minority students may more accurately
reflect true behavioral levels.

Keywords:

ADHD, ethnicity, teacher ratings, direct observations, bias

African American and Hispanic students are

disproportionately diagnosed and placed into
categories of special education in the United
States (Coutinho & Oswald, 2000; Oswald,
Coutinho, Best, & Singh, 1999). Government
education data show overrepresentation of Af-
rican American students in categories of emo-
tional disturbance (ED) and mental retardation
(MR) for each year since 1968 (Donovan &
Cross, 2002; Hosp & Reschly, 2004). Among a
group of one million American students,
160,000 more African Americans students than
Caucasian students will be placed in special
education (Hosp & Reschly, 2004).

Higher rates of identification among ethic

minority students may stem from true cognitive

or behavioral differences, bias in the referral
and assessment process, or some combination
of these two. For the purposes of the current
investigation, “bias” is defined as variation in
teacher ratings of behavior based on student
ethnicity (Chang & Stanley, 2003). If teachers’
perceptions of “typical” and “atypical” behavior
vary across ethnic groups, resulting in different
appraisals or actions based on the same behav-
ior sample, then “bias” is present (Chang &
Stanley, 2003).

Teacher referrals and ratings are central de-

terminants in special education evaluations. Ev-
idence shows referral may represent one source
of placement differences as the majority of stu-
dents referred to special education are ulti-
mately classified and placed (Artiles & Trent,
1994; Ysseldyke, Vanderwood, & Shriner,
1997). Hosp and Reschly (2003) found that 132
African American and 106 Hispanic students
are referred for every 100 Caucasian students.
Studies also show that teacher tolerance is a
primary indicator for identification of behavior
problems and teachers are less tolerant of be-
haviors that are inconsistent with their cultural
expectations (Gerber & Semmel, 1984; Lam-
bert, Puig, Lyubansky, Rowan, & Winfrey,

Shelley J. Hosterman and George J. DuPaul, Department

of Education and Human Services, Lehigh University; and
Asha K. Jitendra, Department of Educational Psychology,
University of Minnesota.

The preparation of this article was supported by NIMH

Grant R01-MH62941.

Correspondence concerning this article should be addressed

to Shelley Hosterman, Lehigh University, Department of Ed-
ucation and Human Services, Iacocca Hall, 111 Research
Drive, Bethlehem, PA 18015. E-mail: [e-mail]

School Psychology Quarterly

Copyright 2008 by the American Psychological Association

2008, Vol. 23, No. 3, 418 – 435

1045-3830/08/$12.00

DOI: 10.1037/a0012668

418

2001). Because 90% of teachers in the U.S. are
Caucasian, it is important to explore the possi-
ble influence of cultural bias on identification of
behavior problems (U.S. Department of Educa-
tion, Office of Educational Research & Im-
provement, 1998).

Attention-deficit/hyperactivity disorder (ADHD)

is among the most common childhood disorders. In
fact, an estimated 3% to 5% of children exhibit
symptoms of ADHD (APA, 2000). Almost 50%
of children with ADHD are eventually placed in
special education programs for behavioral dis-
orders or learning disabilities (Reid, Maag,
Vasa, & Wright, 1994). Students with ADHD
are also assigned to the growing category of
“other health impaired” (National Center for
Education Statistics, 2001).

Evaluations of ADHD symptoms typically

employ a behavioral assessment approach uti-
lizing multiple methods of data collection
across settings and informants (e.g., Anastopou-
los & Shelton, 2000; Barkley, 2006). Major
components include parent and teacher inter-
views, rating scales completed by parents and
teachers, and observations of the child in mul-
tiple settings and task situations (DuPaul &
Stoner, 2003). Thus, responsible diagnosis and
placement decisions rely on a comprehensive
and complex collection of data rather than a
single source (Shapiro & Kratochwill, 2000).

Teacher ratings are a valued aspect of ADHD

assessment because they summarize extensive,
accumulated observations of child behavior
from individuals who are familiar with devel-
opmental expectations (Busse & Beaver, 2000).
These ratings contribute to diagnostic decision-
making by clarifying whether ADHD symp-
toms are inconsistent with developmental level
and associated with impairment across two or
more settings (APA, 2000). Teacher ratings of
behavior are among the most commonly used
methods in school-based assessment of ADHD
(Barkley, 2006; DuPaul & Stoner, 2003). If bias
occurs in teacher ratings of ADHD symptoms,
this may be one source of incongruity in special
education placements across ethnic groups.

Three cross-cultural studies using videotaped

vignettes of scripted ADHD symptoms and dis-
ruptive behaviors have demonstrated the influ-
ence of rater culture on perceptions of ADHD
symptoms. Mann et al. (1992) showed that Chi-
nese and Indonesian clinicians rated hyperac-
tive-disruptive behaviors more severely than

did their Japanese and American colleagues. In
a second study, American and Japanese teachers
rated student behavior more moderately than
did teachers from China, Indonesia, and Thai-
land (Meuller et al., 1995). Similarly, teachers
and student teachers from mainland China
rated videotaped samples of ADHD symp-
toms higher than did teachers from Hong
Kong and the United Kingdom (Alban-
Metcalfe, Cheng-Lai, & Ma, 2002). These
studies emphasize variation in standards for
child behavior across cultures and highlight
the influence of rater’s culture on perception
of ADHD symptoms. A teacher’s personal
history and culture may impact their view of
student behavior.

Other investigations have highlighted the in-

fluence of student ethnicity on teacher percep-
tions of behavior concerns. Stevens (1980)
found teacher ratings of both positive and neg-
ative student characteristics were more influ-
enced by student ethnicity and SES than by
observable behaviors. Results of Prieto and
Zucker (1981) showed that teachers rated a stu-
dent with symptoms of emotional disturbance
as appropriate for services more often when the
student was Mexican American than Caucasian.
Similarly, African American students described
as “difficult to teach” were rated more appro-
priate for special education referral than their
Caucasian peers, regardless of teacher ethnicity
(Bahr, Fuchs, Stecher, & Fuchs, 1991). Middle
school teachers rated students with movement
styles common in African American culture
(e.g., a nonstandard “stroll” style of walking
with a swagger and knees bent) lower in
achievement, higher in aggression, and more in
need of special education than students with
standard movement styles (Neal, McCray,
Webb-Johnson, & Bridgest, 2003). In contrast,
two recent investigations showed that neither
teacher recommendations for placement (Frey,
2002) nor rating scale reports (Cullinan &
Kauffman, 2005) for students with emotional
and behavioral disorders demonstrated bias to-
ward overrepresentation of ethnic minority stu-
dents after SES was controlled. Nonetheless,
evidence suggests student ethnicity may influ-
ence teacher ratings of ADHD symptoms.

Two studies, each examining the correspon-

dence between teacher ratings and direct obser-
vations of varied problem behaviors in cross-
cultural settings, suggested congruence between

419

TEACHER RATINGS OF ADHD IN ETHNIC MINORITY STUDENTS

teacher and student cultures as a key influence.
Puig et al. (1999) compared teacher ratings with
direct observation data in a sample of African
American students and Jamaican students of
African descent. Although Jamaican students
displayed significantly higher levels of observ-
able problem behaviors, teacher ratings of Af-
rican American students were nearly double
those of their Jamaican peers. Puig et al. sug-
gested that Caucasian teachers working in the
U.S. may have lower thresholds of tolerance for
problem behaviors in African American stu-
dents and provide exaggerated reports of these
symptoms. In a second study, direct observa-
tions showed that levels of problem and off-task
behavior in American students were nearly
twice those of Thai students, but reports of
problem behaviors from Thai teachers nearly
doubled those of American teachers (Weisz,
Chaiyasit, Weiss, Eastman, & Jackson, 1995).
Clearly, the match between student and teacher
cultures must be considered when interpreting
rating scale data.

Studies of common ADHD rating scales re-

veal patterns of overidentifying students from
certain ethic minority backgrounds. Initial re-
search in this area emerged after early versions
of scales were developed and interest recently
resurged in response to concerns about cultur-
ally fair assessment and overrepresentation.
Langsdorf, Anderson, Waechter, Madrigal, and
Juarez (1979) examined distributions of scores
on an early version of the Abbreviated Conner’s
Teacher Rating Scale (CTRS; Conners, 1973)
across a large (n

⫽ 1719) sample. Results

showed that in schools with non-Caucasian ma-
jorities, African American students were signif-
icantly more, and Mexican American students
significantly less likely to be rated hyperactive
compared to Caucasian peers. Both SES and
ethnicity had significant effects on CTRS rat-
ings, with poor African American students far-
ing the worst (Langsdorf et al., 1979). Epstein,
March, Conners, and Jackson (1998) found that
African American children were rated higher on
CTRS-R (Conners, 1997) externalizing problem
factors, including conduct disorder and hyper-
activity problems, compared to Caucasian
peers. A third study (Reid, Casat, Norton, Anas-
topoulos, & Temple, 2001) found African
American students would screen positive for
ADHD on the IOWA Conner’s (Pelham,
Milich, Murphy, & Murphy, 1989) at levels

twice those of Caucasian peers. These three
studies provide evidence of potential bias in
teacher ratings of ADHD but cannot distinguish
rater effects from true behavioral variation
across ethnic groups.

The commonly used ADHD-IV Rating

Scale-School Version (DuPaul, Power, Anasto-
poulos, & Reid, 1998) has also been examined
across ethnicity with similar results. In a large
sample of males ages 5 to 18, Reid et al. (1998)
found teachers rated African American students
higher on all symptoms relative to their Cauca-
sian peers. Results showed that screening all
students using norms based on a primarily Cau-
casian standardization sample would lead to
twice as many positive screenings in African
American students. Reid et al. (2000) found an
interaction effect for gender and ethnicity; Af-
rican American males received the most severe
and Caucasian females the least severe ratings.
One additional study (Arnold et al., 2003) found
that African American students scored signifi-
cantly higher on the Swanson, Nolan, and Pel-
ham Rating Scale-IV (SNAP-IV; Swanson,
1992) compared to their Hispanic and Cauca-
sian peers, even after controlling for SES.

The studies discussed above establish that

mean scores on common teacher rating scales of
ADHD vary across ethnic groups. However, the
possibility remains that such ratings may be
accurate reflections of behavioral variation.
Several investigators have suggested comparing
teacher ratings of ADHD symptoms with more
objective direct observations to explore the im-
pact of student ethnicity relative to true behav-
ior difference (Epstein et al., 1998; Langsdorf
et al., 1979; Reid et al., 2001, 1998). Only three
studies in the literature have used this method to
investigate the possibility of ethnic bias in
teacher ratings of ADHD symptoms.

Ramirez and Shapiro (2005) examined the

effects of both teacher and student ethnicity on
ratings of ADHD symptoms. Caucasian and
Hispanic teachers completed ADHD-IV rating
scales after viewing videotaped vignettes of stu-
dents exhibiting similar levels of overactive,
inattentive, or disruptive behaviors. Results
showed effects of teacher ethnicity, specifically
that Hispanic teachers may hold students of
their own culture to stricter standards of behav-
ior. Student ethnicity had no effect on the rat-
ings of Caucasian teachers in this analog situa-
tion. Although this study provided excellent

420

HOSTERMAN, D

U

PAUL, AND JITENDRA

controls of ethnicity and behavior, it may not
accurately represent the use of ADHD rating
scales in school settings, where teachers typi-
cally base behavioral ratings on accumulated
interaction with students rather than discrete
samples.

Working in the United Kingdom, Sonuga-

Barke, Minocha, Taylor, and Sandberg (1993)
compared teacher reports from interviews and
questionnaires with direct observations and me-
chanical data on student behavior in analog and
regular classroom settings. Teacher reported hy-
peractivity levels of Asian (parents from India,
Pakistan, or Bangladesh) students were overes-
timates as compared to objective measures of
student behavior. In fact, direct observations
showed the behavior of Asian students rated
“hyperactive” was comparable to that of En-
glish students in the control group. This study
suggests that comparisons of subjective teacher
ratings to more objective systematic observa-
tions may reveal teacher bias toward certain
ethnic groups (Sonuga-Barke et al., 1993).

The third study, Epstein et al. (2005), com-

pared teacher ratings on three common scales to
observation data of African American and Cau-
casian students. Results showed that overall,
teacher ratings of ADHD symptoms were
higher for the African American group (Epstein
et al., 2005). However, direct observation data
revealed a similar pattern, with African Amer-
ican students exhibiting higher levels of observ-
able ADHD symptoms than their Caucasian
peers (Epstein et al., 2005). Correlations of
teacher ratings and observation data were sim-
ilar for both groups. Thus, unlike Sonuga-Barke
et al. (1993), these results did not evidence bias
in teacher ratings of ADHD symptoms for stu-
dents from ethnic minority backgrounds.

Clearly, the literature examining the question

of bias in teacher ratings of ADHD symptoms
by comparing teacher rating scale data with
direct observations of classroom behavior is
both sparse and inconclusive. The purpose of
the current study was to expand this literature
by comparing teacher ratings on two commonly
used ADHD ratings scales to direct observation
data from a well-established observational code.
The study aimed to extend the work of Sonuga-
Barke et al. (1993) to a sample of American
students. Also, it extended the U.S. based work
of Epstein et al. (2005) by including a sample

with an ethnic composition more representative
of the U.S. population.

The current study aimed to answer the ques-

tion: Are teacher ratings of ADHD symptoms in
African American and Hispanic students less
consistent with objective observations than are
teacher ratings of Caucasian students? The hy-
pothesis was that teacher ratings of ADHD
symptoms would correlate more strongly with
behavioral observations for Caucasian students
than for a group comprised of African American
and Hispanic students.

Method

Participants

Sample of Larger Investigation

Participants were selected from a larger study

examining academic interventions for elemen-
tary school students with ADHD (for details,
see DuPaul, Jitendra et al., 2006). All partici-
pants met several initial inclusion criteria. Re-
cruited from schools in eastern Pennsylvania,
all students were enrolled in a first through
fourth grade, general or special education class-
room in the public school system. Participants’
families also indicated that they did not plan to
move from the area within 2 years following the
start of the study.

Students in the ADHD group met four addi-

tional inclusion criteria. Children were consid-
ered for the ADHD group if they either had a
previous ADHD diagnosis from an outside ser-
vice provider or were reported by a teacher to
exhibit significant problems with inattention,
impulsivity and/or hyperactivity. All potential
participants were then screened based on the
following criteria. All students in the ADHD
group scored at or above the 90th percentile on
one or both of the Inattention or Hyperactivity-
Impulsivity subscales of the ADHD-IV (DuPaul
et al., 1998) across both parent and teacher
ratings. Higher ratings and percentiles on the
ADHD-IV represent more severe ADHD symp-
toms. Each student also met DSM–IV (APA,
2000) criteria for one of the three ADHD sub-
types based on a parent interview with the
NIMH Diagnostic Interview Schedule for Chil-
dren-IV (DISC-IV; Shaffer, Fisher, & Lucas,
1998), administered by trained graduate stu-
dents. This interview was used to ensure that

421

TEACHER RATINGS OF ADHD IN ETHNIC MINORITY STUDENTS

students met a sufficient number of individual
symptoms to warrant diagnosis. Finally, each
child was also experiencing achievement prob-
lems in either math or reading, according to
teacher report. Inclusion in the ADHD group
was based on reports from multiple respondents
and settings. These criteria addressed symptom
severity by using stringent cut-points on the
ADHD-IV and ensuring that participants met
DSM–IV criteria for ADHD.

Students in the non-ADHD control group

were recruited from the same schools, but from
different classrooms than children in the ADHD
group. Their teachers indicated that these stu-
dents were experiencing no behavioral or aca-
demic problems. Each student in the control
group scored below the 90th percentile on all
subscales of the ADHD-IV (DuPaul et al.,
1998) and did not meet criteria for ADHD,
Oppositional Defiant Disorder (ODD), or Con-
duct Disorder (CD) based on the DISC-IV par-
ent version (Shaffer, Fisher, & Lucas, 1998).
These students were also matched to those in
the ADHD group by grade and gender.

For the purpose of the current investigation,

all Caucasian, African American, and Hispanic
participants from both the ADHD and non-
ADHD control groups of the larger study were
considered for inclusion. That is, the current
study considered students classified into both
the ADHD (n

⫽ 175) and non-ADHD control

(n

⫽ 66) groups of the larger investigation for

inclusion, working from a combined original
sample of 241 students. The sample of the
larger study represented the ethnic composition
of the geographic area and included students
classified as Caucasian (n

⫽ 146), African

American (n

⫽ 22), Hispanic (n ⫽ 64), and

other (n

⫽ 6).

Sample of Current Investigation

Using this sample, two groups were created:

a Caucasian group and an ethnic minority group
comprised of all African American and His-
panic students in the sample. Inclusion was fur-
ther limited to ensure independence of raters.
That is, if a classroom teacher had two or more
students enrolled in the larger study, one student
was randomly selected for the current investi-
gation. Inclusion was also limited by teacher
completion of behavior rating scales. For exam-
ple, data obtained from a direct observation in
math were included only if the student’s math
teacher completed the behavior ratings scales.
Because separate analyses were conducted for
reading and math, the same student could be
represented in analyses for both academic areas.

Table 1 displays demographic data for the

final sample of the current study. These data are
shown separately for the Caucasian (n

⫽ 112)

and ethnic minority (n

⫽ 60) groups. The ethnic

minority group consisted of 17 (28.3%) African
American, 38 (63.3%) Hispanic, and 5 (8.3%)
students of both African American and Hispanic
descent. The Caucasian group was 75% (n

⫽

85) male and the ethnic minority Group 90%
male (n

⫽ 54). At enrollment, 21% of partici-

pants were taking psychotropic medication (pri-
marily psychostimulants). Based on results of
the DISC-IV (Shaffer et al., 1998), 17.7% of
participants met criteria for ADHD: Inattentive
subtype, 6.9% for ADHD Hyperactive-Impul-
sive subtype, 46% for the combined subtype,
and the remainder were control students.

Among the 171 students included in the reading

analyses, 54 (31.6%) displayed academic difficul-
ties in reading only, 35 experienced difficulty in
both reading and math (20.4%), and 82 (48%)

Table 1
Demographic Characteristics of the Caucasian and Ethnic Minority Groups

Measure

Group

Caucasian (n

⫽ 112)

Ethnic minority (n

⫽ 60)

t(170) or

␹

2

(1)

p

Age (in years)

8.48

a

(1.25)

b

8.49 (1.13)

⫺.052

0.959

Male (%)

75

90

5.543

.019

ⴱ

SES

59.71 (22.62)

43.62 (26.90)

⫺2.596

.010

ⴱ

Group (% ADHD group)

69.6

76.7

.958

.328

Note.

SES

⫽ social economic status determined by highest Hollingshead parent occupation; ADHD ⫽ attention deficit

hyperactivity disorder.

a

Mean.

b

Standard deviation.

ⴱ

p

⬍ .05.

422

HOSTERMAN, D

U

PAUL, AND JITENDRA

displayed difficulty in either math alone or in
neither academic area. Among the 167 students
included in the math analyses, 26 (15.6%) had
difficulties in math only, 35 (21.0%) had diffi-
culties in both reading and math, and 106
(63.5%) had difficulties either exclusively in
reading or in neither academic area.

Among students receiving academic inter-

ventions, teacher ratings of academic perfor-
mance in reading and mathematics on the Aca-
demic Competence Evaluation Scales (ACES;
DiPerna & Elliott, 2000) shared significant cor-
relations with student performance on reading
and math subtests of the Woodcock-Johnson III
Test of Achievement (WJ-III; Woodcock,
McGrew, & Mather, 2001). Teacher ratings of
student reading ability on the ACES accounted
for 19.8% of variance (r

⫽ .445, p ⬍ .001) in

standard scores on the WJ-III Reading Fluency
subscale and 22.8% of variance (r

⫽ .477, p ⬍

.001) in standard scores on the WJ-III Passage
Comprehension subscale. Teacher ratings of
mathematics ability on the ACES accounted
for 11.4% of variance (r

⫽ .338, p ⫽ .003) in

standard scores on the WJ-III Calculation
subtest and 10.5% of variance (r

⫽ .324, p ⫽

.005) in standard scores on the Math Fluency
subtest. At baseline, students in the math inter-
vention group earned standard scores in the low
average range on the Calculation (M

⫽ 92.15,

SD

⫽ 13.81) and Math Fluency (M ⫽ 86.66,

SD

⫽ 13.39) subtests of the WJ-III. Students

receiving reading intervention scored in the low
average range on the Reading Fluency
(M

⫽ 81.33, SD ⫽ 21.04) and Reading Com-

prehension (M

⫽ 88.30, SD ⫽ 11.56) subtests

of the WJ-III. Independent sample t tests re-
vealed no significant differences between the
Caucasian and ethnic minority groups on WJ-III
scores or ACES ratings.

Teacher demographic data were only avail-

able for teachers of students in the ADHD group
of the larger study (70% of included cases). Of
these teachers (n

⫽ 120), 92% were fe-

male, 95.5% were Caucasian, 90.9% taught in
regular education classrooms, and 47.7% pos-
sessed a master’s degree. Teachers aver-
aged 10.90 (SD

⫽ 9.13) years of experience.

Demographics of teachers with students in the
control group are expected to be similar, given
that they were teaching in the same school settings
as those in the ADHD group. Because teacher
referral and ratings were part of the screening

process, teachers were not blind to group mem-
bership or purpose of the larger study. However,
teachers were unaware that questions related to
student ethnicity would be investigated.

Procedures

Data for the current investigation came from

the initial (pretreatment) collection phase of the
larger study. Trained graduate students col-
lected direct observation data for each partici-
pant during 15 minutes of both a regular reading
and regular math class. All participants were
observed in both reading and math, regardless
of academic performance. Observations typi-
cally occurred during January or February. Ob-
servers were kept blind to group membership
(ADHD vs. non-ADHD control) in the larger
study. A significant portion of all observations
were collected by two raters simultaneously to
monitor interobserver agreement (IOA). Occur-
rence, nonoccurrence, and overall IOA values
were calculated by dividing the number of
agreements by the sum of the number of agree-
ments plus the number of disagreements, and
multiplying this value by 100%. Kappa values
were calculated to examine the level of agree-
ment beyond chance.

In math, IOA data were available for 27.5% of

observations. Reported in the order of occurrence,
nonoccurrence, total, and kappa, IOA values for
each observational code in math were as follows:
Active Engaged Time (0.95, 0.96, 0.98, 0.95),
Passive Engaged Time (0.94, 0.96, 0.98, 0.94),
Off-Task Motor (0.91, 0.97, 0.98, 0.93), Off-
Task Verbal (0.88, 0.99, 0.99, 0.90), and Off-
Task Passive (0.90, 0.99, 0.99, 0.93). For read-
ing observations, paired IOA observations were
collected for 31% of all observations. IOA val-
ues for each code in reading were as follows:
Active Engaged Time (0.95, 0.98, 0.98, 0.96),
Passive Engaged Time (0.93, 0.94, 0.97, 0.93),
Off-Task Motor (0.89, 0.98, 0.98, 0.92), Off-
Task Verbal (0.89, 0.99, 0.99, 0.91), and Off-
Task Passive (0.85, 0.99, 0.99, 0.89).

Each student’s teacher also completed the

two behavior rating scales detailed in the
measures section. Teachers did not receive
explicit training on completion of rating
scales but were provided with a contact num-
ber for any questions or clarifications. It was
expected that teachers would appropriately
follow the standardized written instructions

423

TEACHER RATINGS OF ADHD IN ETHNIC MINORITY STUDENTS

on each scale. Forms were given to teachers
on the day of the direct observation with
instructions to return ratings promptly. How-
ever, the large sample size and geographic
area of the study impacted the latency be-
tween observations and return of rating
scales. Scales were returned at teacher con-
venience or after a reminder call but were
typically received within 4 to 8 weeks of the
observation.

Measures

SES

SES was measured using the parental occu-

pation portion of the 4-factor index of social
status (Hollingshead, 1976), a common measure
of SES. This measure collects participant gen-
erated, open-ended descriptions of each parent’s
occupation. Occupations were translated into
values between 0 and 90, where higher numbers
indicate more financially lucrative occupations
and a higher overall level of SES. For the pur-
pose of the current study, SES was represented
by selecting the parental occupation (e.g., either
mother or father) with the highest value on the
Hollingshead list. The full range of scores rep-
resenting parental occupation (rather than col-
lapsed categories of similar occupations) was
used to capture the complete range of variability
in SES.

CTRS-R:L

The CTRS-R:L (Conners, 1997) is a common

teacher rating assessment of behavior problems
in the classroom. The scale has two indexes,
DSM–IV: Inattentive and DSM–IV: Hyperac-
tive-Impulsive, each comprised of 9, 4-point
Likert style items. High scores on each index
correspond with DSM–IV diagnostic criteria for
the ADHD subtypes. The DSM–IV: Inattentive
Index has internal consistency from 0.87 to 0.96
across age ranges and gender, and test–retest
reliability (over 6 to 8 weeks) of 0.70. For the
DSM–IV: Hyperactive-Impulsive Index, inter-
nal consistency is between 0.82 and 0.95, and
test–retest reliability (over 6 to 8 weeks) is 0.47
(Conners). The CTRS-R:L provides T scores
based on age and gender.

ADHD-IV School Version

The ADHD-IV School Version (DuPaul et

al., 1998) is adapted from the ADHD symptom
list specified in the DSM–IV (APA, 2000) and
consists of 18, 4-point Likert style items. The
two, 9-item subscales, Inattention and Hyperac-
tivity-Impulsivity, correspond to DSM–IV diag-
nostic categories. Each subscale has excellent
psychometric properties. The Inattention scale
has an internal consistency coefficient of 0.96
and a 4-week test–retest reliability of 0.89. The
Hyperactivity-Impulsivity scale has an internal
consistency coefficient of 0.88 and a 4-week
test–retest reliability of 0.88. The ADHD-IV
does not provide T scores, thus raw scores were
used in this study.

Behavioral Observation System for
Schools (BOSS)

The BOSS (Shapiro, 2003) provides direct

observations of behavioral symptoms of ADHD
in the classroom setting. Observations are seg-
mented into 15-sec intervals and four randomly
selected classroom peers provide periodic com-
parison data. Five BOSS behavior codes were
used for the current study: Active Engaged Time
(AET), Passive Engaged Time (PET), Off Task
Motor (OFT-M), Off Task Verbal (OFT-V), and
Off Task Passive (OFT-P). Two of the behaviors,
AET and PET, are recorded through momentary
time sampling. The remaining three codes are
observed by partial interval recording. Each be-
havior is reported by the percentage of intervals in
which it is observed.

Data Analyses

Chi-square tests and independent samples

t tests were conducted on key demographic vari-
ables (gender, SES, age) to uncover any be-
tween group differences. Any statistically sig-
nificant differences in these variables across
groups were then accounted for by treating the
variable as a covariate in the final analyses.

Separate analyses were computed with data

from reading class and math class. Thus, each of
the analyses described below was conducted
separately for each of the two academic areas.
Pearson correlation coefficients were calculated
between the hyperactivity-impulsive indexes of
both the CTRS-R:L (Conners, 1997) and the

424

HOSTERMAN, D

U

PAUL, AND JITENDRA

ADHD-IV School version (DuPaul et al., 1998)
and the Off-Task Motor and Off-Task Verbal
behaviors off the BOSS (Shapiro, 2003). A sec-
ond set of Pearson correlation coefficients was
computed between the inattentive indexes of
both the CTRS-R:L and the ADHD-IV School
version and each of the following BOSS codes:
Active Engaged Time, Passive Engaged Time,
and Off-Task Passive.

In order to obtain power of 0.80 for a large

effect size (assuming an alpha level of 0.05) in
the selected data analyses, each group (Cauca-
sian and ethnic minority) must contain 33 stu-
dents (n

⫽ 66). Thus, the current sample was

sufficient to detect large effects.

Finally, comparisons of corresponding corre-

lation coefficients from the Caucasian and eth-
nic minority groups were made to determine if
accuracy in teacher ratings of ADHD symptoms
differed based on student ethnicity. To make
this comparison, all correlation coefficients
were first transformed with a Fisher’s z

⬘ Trans-

formation (Cohen & Cohen, 1983). Values for
Fisher’s r to z

⬘ transformation were obtained

directly from a standard table for this conver-
sion (Appendix Table B; Cohen & Cohen,
1983). With these z

⬘ values, the normal curve

deviate was computed using formula 2.8.5 from
Cohen and Cohen. The normal curve deviate
provides a test of the significance of the differ-
ence between correlation coefficients obtained
on two independent samples (Cohen & Cohen,
1983). The generated z-scores were then eval-
uated for significance using a Standard Normal
Distribution table (e.g., Appendix Table C.; Co-
hen & Cohen, 1983). Values meeting the two-
tailed

␣ ⫽ .05 criteria were considered signifi-

cant differences.

In addition to evaluating statistically signifi-

cant difference, effect sizes for differences in
corresponding correlation coefficients were also
computed as indicated in Cohen (1988). That is,
the effect size value of q

⫽ z

1

⫺ z

2

, where z

n

is

the Fisher’s z

⬘ transformation of the relevant

correlation coefficient, was used. Table 4.2.2
was used to covert each r to z

⬘ via Fisher’s z⬘

Transformation. Next, the absolute difference
(q) between each corresponding pair of z

⬘ val-

ues was compared against Cohen’s standards
for small (q

⫽ .10), medium (q ⫽ .30), and

large (q

⫽ .50) effect sizes.

Results

Demographic Comparisons

Results of chi-square tests and independent

samples t test comparisons revealed statistically
significant differences in both gender and SES
between the Caucasian and ethnic minority
groups. The ethnic minority group contained a
significantly higher proportion of male students,

␹

2

(1)

⫽ 5.543, p ⫽ .019, than did the Caucasian

group. Students in the Caucasian group
(M

⫽ 59.71) came from households with sig-

nificantly higher levels of SES, t(170)

⫽

⫺2.596, p ⫽ .01, as compared to students in the
ethnic minority group (M

⫽ 43.62). However,

the two groups were equivalent in both age,
t(170)

⫽ ⫺.052, p ⫽ .959, and proportions of

students belonging to the ADHD and control
groups of the larger study,

␹

2

(1)

⫽ .958, p ⫽

.328. Significant differences between groups
were addressed using covariates and T scores
during later analyses.

Descriptive Statistics

Descriptive statistics for all indicators are

summarized in Table 2 along with t test com-
parisons between the Caucasian and ethnic mi-
nority groups. Only one statistically significant
between-groups difference occurred in direct
observation (BOSS) data. Specifically, com-
pared to students in the Caucasian group, stu-
dents in the ethnic minority group displayed
significantly higher levels of off-task verbal be-
havior during direct observations in both read-
ing, t(169)

⫽ ⫺2.818, p ⫽ .006, and math,

t(166)

⫽ ⫺2.28, p ⫽ .025, classes. Levels of the

remaining direct observation behaviors were
equivalent across groups in both reading and
math settings.

Values on one of the four teacher rating

scale indicators differed significantly between
groups. Teacher ratings on the ADHD-IV Hy-
peractive-Impulsive Index were significantly
higher for students in the ethnic minority
group in both reading, t(155)

⫽ ⫺2.368, p ⫽

.019, and math, t(153)

⫽ ⫺2.177, p ⫽ .031.

No significant differences between groups
were uncovered on the three remaining
teacher indicators in either reading or math.

425

TEACHER RATINGS OF ADHD IN ETHNIC MINORITY STUDENTS

Table

2

Descriptive

and

Comparison

Statistics

Across

Groups

Measure

Reading

Math

BOSS

data

(%

intervals)

Caucasian

Ethnic

minority

t(169)

p

Caucasian

Ethnic

minority

t(166)

p

OFT-M

15.1%

a

(16.0)

b

17.5%

(16.9)

⫺

.922

.358

19.7%

(19.1)

20.7%

(17.6)

⫺

.338

.736

OFT-V

5.0%

(7.2)

9.4%

(10.9)

⫺

2.818

.006

ⴱ

6.2%

(8.6)

10.2%

(11.9)

⫺

2.28

.025

ⴱ

OFT-P

7.4%

(8.4)

7.2%

(8.0)

.157

.875

7.7%

(8.2)

6.9%

(7.4)

.656

.513

AET

26.6%

(16.1)

24.9%

(18.9)

.624

.533

28.5%

(17.2)

29.9%

(14.7)

⫺

.274

.785

PET

49.8%

(19.3)

45.9%

(23.1)

1.106

.271

44.0%

(20.2)

38.8%

(18.7)

1.612

.109

Reading

Math

Teacher

rating

scale

data

Caucasian

Ethnic

minority

t(155)

p

Caucasian

Ethnic

minority

t(153)

p

CTRS:Hyperactive-Impulsive

59.80

c

(14.06)

63.56

(14.26)

⫺

1.589

.114

59.66

(14.07)

63.23

(14.19)

⫺

1.501

.135

ADHD

⫺

IV:Hyperactive-Impulsive

11.03

d

(9.01)

14.75

(9.91)

⫺

2.368

.019

ⴱ

11.06

(9.08)

14.52

(9.86)

⫺

2.177

.031

ⴱ

CTRS:Inattentive

61.79

c

(13.96)

61.83

(12.24)

⫺

.019

.985

61.55

(13.97)

61.58

(12.22)

⫺

.015

.988

ADHD

⫺

IV:Inattentive

15.28

d

(10.30)

16.62

(9.60)

⫺

.791

.430

15.23

(10.34)

16.42

(9.58)

⫺

.693

.489

Note.

BOSS

⫽

behavioral

observation

of

students

in

the

schools

code;

OFT-M

⫽

off-task

motor

behavior;

OFT-V

⫽

off-task

verbal

behavior;

OFT-P

⫽

off-task

passive

behavior;

AET

⫽

active

engaged

time;

PET

⫽

passive

engaged

time;

CTRS

⫽

Conners

Teacher

Rating

Scale

Revised:

Long

Form;

ADHD-IV

⫽

ADHD-IV

School

Version.

a

Mean.

b

Standard

deviation.

c

T-score.

d

Raw

score.

ⴱ

p

⬍

.05.

426

HOSTERMAN, D

U

PAUL, AND JITENDRA

Intercorrelations Between Direct

Observation Data and Teacher Ratings

Table 3 displays Pearson correlation coef-

ficients between each direct observation be-
havior and the corresponding teacher rating
scale indicators. Correlations are presented
separately by group and academic subject.
Because both gender and SES differed signif-
icantly between the Caucasian and ethnic mi-
nority groups, it was necessary to control for
these two variables in all correlations. For
CTRS-R:L indicators, T score values (ad-
dressing gender) were used within partial cor-

relations controlling for SES. Because T
scores are not available for the ADHD-IV,
raw scores were used within partial correla-
tions controlling for both SES and gender.

Intercorrelations in Reading Data

Results indicated several statistically signifi-

cant associations between direct observations in
reading and teacher ratings of ADHD symp-
toms. For the Caucasian group, off-task motor
behavior accounted for 7.6% of variance on the
CTRS-R:L DSM–IV: Hyperactive-Impulsive in-
dex (r

⫽ .276, p ⫽ .006) and 12.7% of variance

Table 3
Intercorrelations Between Observed Behaviors and Teacher Rating Indexes

Reading

CTRS: Hyperactive-

Impulsive

a

ADHD-IV: Hyperactive-

Impulsive

b

CTRS: Inattentive

a

ADHD-IV: Inattentive

b

Caucasian students (n

⫽ 98)

OFT-M

.276

ⴱ

.356

ⴱⴱ

—

—

OFT-V

.092

.111

—

—

AET

—

—

.014

.000

PET

—

—

⫺.271

ⴱ

⫺.254

ⴱ

OFT-P

—

—

.246

ⴱ

.266

ⴱ

Ethnic minority students (n

⫽ 43)

OFT-M

.438

ⴱⴱ

.434

ⴱ

—

—

OFT-V

.231

.237

—

—

AET

—

—

⫺.330

ⴱ

⫺.350

ⴱ

PET

—

—

.055

.043

OFT-P

—

—

.252

.300

Math

CTRS: Hyperactive-

Impulsive

ADHD-IV: Hyperactive-

Impulsive

CTRS: Inattentive

ADHD-IV: Inattentive

Caucasian students (n

⫽ 99)

OFT-M

.254

ⴱ

.293

ⴱⴱ

—

—

OFT-V

.068

.108

—

—

AET

—

—

⫺.099

⫺.114

PET

—

—

⫺.121

⫺.077

OFT-P

—

—

.172

.208

ⴱ

Ethnic minority students (n

⫽ 43)

OFT-M

.413

ⴱ

.491

ⴱⴱ

—

—

OFT-V

.470

ⴱⴱ

.556

ⴱⴱ

—

—

AET

—

—

⫺.219

⫺.267

PET

—

—

⫺.390

ⴱ

⫺.462

ⴱⴱ

OFT-P

—

—

.383

ⴱ

.399

ⴱ

Note.

BOSS

⫽ behavioral observation of students in the schools code; OFT-M ⫽ off-task motor behavior; OFT-V ⫽

off-task verbal behavior; OFT-P

⫽ off-task passive behavior; AET ⫽ active engaged time; PET ⫽ passive engaged time;

CTRS

⫽ Conners Teacher Rating Scale Revised: Long Form; ADHD-IV ⫽ ADHD-IV School Version; ADHD-IV ⫽

ADHD-IV School Version.

a

Partial correlations controlling for SES and using T-scores on CTRS.

b

Partial correlations

controlling for SES and gender and using raw scores on ADHD-IV.

ⴱ

p

⬍ .05;

ⴱⴱ

p

⬍ .005.

427

TEACHER RATINGS OF ADHD IN ETHNIC MINORITY STUDENTS

on the ADHD-IV Hyperactive-Impulsive index
(r

⫽ .356, p ⬍ .001). The same correlations

were also significant in the ethnic minority
group, with off-task motor behavior accounting
for 19.2% of variance in CTRS-R:L DSM–IV:
Hyperactive-Impulsive ratings (r

⫽ .438, p ⫽

.004) and 18.8% of variance in ratings on the
ADHD-IV Hyperactive-Impulsive index (r

⫽

.434, p

⫽ .005).

In contrast, teacher ratings of Inattentive

symptoms related to distinct sets of direct
observation behaviors for each of the two
groups. For the Caucasian group, passive en-
gaged time (r

⫽ ⫺.271, p ⫽ .007) accounted

for 7.3% of variance in the CTRS-R:L DSM–
IV: Inattentive scale, and off-task passive be-
havior (r

⫽ .246, p ⫽ .015) accounted for 6%

of the variance in this scale. Similarly, pas-
sive engaged time accounted for 6.5% of the
variance in ratings on the ADHD-IV Inatten-
tive scale (r

⫽ ⫺.254, p ⫽ .013) and off-task

passive behavior accounted for 7% of the
variance in this scale (r

⫽ .266, p ⫽ .009) for

the Caucasian group.

None of these associations were significant

in the ethnic minority group. For this group,
active engaged time was the only variable
significantly associated with ratings of Inat-
tentive behavior. Active engaged time ac-
counted for 10.9% of variance in teacher rat-
ings on the CTRS-R:L DSM–IV: Inattentive
scale (r

⫽ ⫺.330, p ⫽ .033) and 12.3% of

variance in ratings on the ADHD-IV Inatten-
tive scale (r

⫽ ⫺.350, p ⫽ .025).

Intercorrelations in Math Data

In math analyses, a greater proportion of the

hypothesized correlations were statistically signif-
icant for the ethnic minority group as compared to
the Caucasian group. Only 3 of the 10 hypothe-
sized associations were significant in the latter
group. As in the reading analyses, off-task motor
behavior was significantly correlated with teacher
ratings of hyperactive behavior in both groups.
For the Caucasian group, off task motor behavior
accounted for 6.5% of variance on the CTRS-R:L
DSM–IV: Hyperactive-Impulsive scale (r

⫽ .254,

p

⫽ .011) and 8.6% of variance on the ADHD-IV

Hyperactive-Impulsive scale (r

⫽ .293, p ⫽ .004).

Only one additional correlation was significant for
the Caucasian group. Specifically, off-task passive

behavior accounted for 4.3% of variance in the
ADHD-IV Inattentive scale (r

⫽ .208, p ⫽ .043).

In the ethnic minority group, 8 of the 10

hypothesized correlations between direct obser-
vation data and teacher ratings were significant.
Off task motor behavior accounted for 17.1% of
variance on the CTRS-R:L DSM–IV: Hyperac-
tive-Impulsive scale (r

⫽ .413, p ⫽ .007)

and 24.1% of variance on the ADHD-IV Hy-
peractive-Impulsive scale (r

⫽ .491, p ⫽ .001).

Off-task verbal behavior was also significantly
correlated with these scales, accounting
for 22.1% of variance on the CTRS-R:L DSM–
IV: Hyperactive-Impulsive scale (r

⫽ .470, p ⫽

.002) and 30.9% of variance on the ADHD-IV
Hyperactive-Impulsive scale (r

⫽ .556, p ⬍

.001).

Both passive engaged time and off-task

passive time were significantly correlated
with teacher ratings of Inattentive behavior in
the ethnic minority group. Passive engaged
time accounted for 15.2% of variance on the
CTRS-R:L DSM–IV: Inattentive scale (r

⫽

⫺.390, p ⫽ .011) and 21.3% of variance on
the ADHD-IV Inattentive scale (r

⫽ ⫺.462,

p

⫽ .002). Finally, for the ethnic minority

group, off-task passive behavior was signifi-
cantly correlated with the CTRS-R:L DSM–
IV: Inattentive scale (r

⫽ .383, p ⫽ .012) and

the ADHD-IV Inattentive scale (r

⫽ .399,

p

⫽ .01) accounting for 14.7% and 15.9% of

variance in the two ratings respectively.

Differences in Strength of Correlations

Between Groups

The strength of corresponding correlation co-

efficients from the Caucasian and ethnic minor-
ity groups were compared using both tests of
statistical significant and effect sizes. For statis-
tical comparisons, an online calculator (Univer-
sity of Amsterdam Faculty of Humanities, n.d.),
which automatically completes the transforma-
tions and comparisons detailed in the methods
section, was used to make these comparisons.
Results revealed several pairs of correlations
that differed significantly between the Cauca-
sian and ethnic minority groups.

In math, the positive correlation between off-

task verbal behavior and the CTRS-R:L DSM–
IV: Hyperactive-Impulsive scale was signifi-
cantly higher ( p

⫽ .02) for the ethnic minority

group (r

⫽ .470) as compared to the Caucasian

428

HOSTERMAN, D

U

PAUL, AND JITENDRA

group (r

⫽ .068). Similarly, correlations be-

tween off-task verbal behavior and the
ADHD-IV Hyperactive-Impulsive scale were
significantly higher ( p

⫽ .009) in the ethnic

minority group (r

⫽ .556) than the Caucasian

group (r

⫽ .108). Finally, negative correlations

between passive engaged time and the
ADHD-IV Inattentive scale were significantly
larger ( p

⫽ .03) for the ethnic minority (r ⫽

⫺.462) group when compared to the Caucasian
group (r

⫽ ⫺.077). Other correlation pairs

showed a similar (but nonsignificant) pattern
between groups in the math analyses. In fact,
the magnitude of all 10 hypothesized correla-
tions of direct observation behavior and teacher
ratings in math was greater in the ethnic minor-
ity group as compared to the Caucasian group.
However, none of these differences were statis-
tically significant. In reading analyses, compar-
isons revealed no statistically significant differ-
ences in corresponding correlations between the
two groups.

Effect sizes were computed to measure the

magnitude of difference in corresponding correla-
tions between the ethnic minority and Caucasian
groups. Effect sizes q

⫽ z

1

⫺ z

2

were computed

by subtracting the Fisher’s z

⬘ transformation for

the Caucasian group (z

2

) from the corresponding

value for the ethnic minority group (z

1

). Thus, a

positive effect size indicates a more positive asso-
ciation between two variables within the ethnic
minority group as compared to the Caucasian
group. Similarly, a negative q value indicates a
stronger negative correlation in the ethnic minor-
ity group relative to the Caucasian group.

In reading, three medium (

兩q兩 ⱖ .30) effects

were observed. Medium effect sizes for ethnic-
ity were found when comparing the strength of
correlations between active engaged time and
both the CTRS-R:L: Inattentive subscale (q

⫽

⫺.353) and ADHD-IV: Inattentive subscale
(q

⫽ ⫺.365). Significant negative correlations

between active engaged behavior and ratings
of inattention were observed in the ethnic minority
group (r

⫽ ⫺.33 to ⫺.35), but corresponding

correlations for Caucasian students were near zero
(r

⫽ 0 to .014). A medium effect (q ⫽ .337) for

ethnicity was also found between correlations of
passive engaged time and the CTRS-R:L Inatten-
tive subscale. That is, passive engaged behaviors
shared a significant negative correlation with rat-
ings of inattention for Caucasian students (r

⫽

⫺.271) but were not significantly correlated (r ⫽

.055) with teacher ratings on the CTRS-R:L: In-
attentive subscale for ethnic minority students.
Effect sizes for the remaining reading compari-
sons were either small or negligible.

In math, a large effect size (q

⫽ .523) for

ethnicity was found for the difference in mag-
nitude of correlations between off-task verbal
behavior and the CTRS-R:L: Hyperactive-
Impulsive scale. These variables shared a sig-
nificant positive correlation (r

⫽ .470) for eth-

nic minority students, but were not significantly
correlated (r

⫽ .068) for Caucasian students.

Similarly, a medium effect size (q

⫽ .44) for

ethnicity emerged in comparisons of correla-
tions between off-task verbal behavior and rat-
ings on the ADHD-IV: Hyperactive-Impulsive
scale between the ethnic minority (r

⫽ .47) and

Caucasian (r

⫽ .068) groups. Comparisons of

correlations between both active engaged and
passive engaged time and the ADHD-IV: Inat-
tentive subscale revealed a medium effect sizes
(q

⫽ ⫺.387 and q ⫽ ⫺.427) for ethnicity with

significant negative correlations for ethnic mi-
nority students (r

⫽ ⫺.267 and r ⫽ ⫺.462)

compared to nonsignificant correlations for
Caucasian students (r

⫽ ⫺.114 and r ⫽ ⫺.077).

Finally, a medium effect (q

⫽ ⫺.291) for eth-

nicity was observed when the magnitude of
correlations between passive engaged time and
the CTRS-R:L: Inattentive subscale was com-
pared across ethnic minority (r

⫽ ⫺.39) and

Caucasian (r

⫽ ⫺.077) groups.

1

Discussion

The current study explored how associa-

tions between teacher ratings of ADHD
symptoms and direct behavioral observations
differ in ethnic minority and Caucasian stu-
dent groups. Results showed significant vari-
ation in this association across student ethnic-
ity. However, the influence of student ethnicity
was contrary to prediction. Teacher ratings re-
flecting ADHD symptoms and on-task behavior

1

Effect size data for magnitude of differences in corre-

lations between individual ethnic groups (e.g., Caucasian,
Hispanic, and African American) were computed as post
hoc analyses. Results were similar to those of the planned
analyses. That is, correlations between observational data and
teacher ratings were stronger in both African-American and
Hispanic sub-samples relative to the Caucasian sample.
Results of these analyses did not change the general con-
clusions of this investigation.

429

TEACHER RATINGS OF ADHD IN ETHNIC MINORITY STUDENTS

in ethnic minority students were more consistent
with direct observation data than were ratings of
Caucasian students. These results suggest
teacher ratings of both hyperactive-impulsive
and inattentive symptoms may more accurately
reflect directly observable behaviors in ethnic
minority students as opposed to Caucasian stu-
dents. It appears teacher reports on the CTRS-
R:L and ADHD-IV may be more sensitive to
“true” behavior levels of ethnic minority stu-
dents than those of their Caucasian peers. Prior
studies in this area have focused on issues of
true behavioral difference, influence of student
ethnicity, rater effects, and instrument bias.
These four concepts will guide discussion and
explanation of the current results.

Observable Behavioral Difference

Comparison of teacher ratings must consider

the possibility of true behavioral difference. Ep-
stein et al. (2005) found teachers’ elevated rat-
ings of ADHD symptoms in African American
students were accurate reflections of directly
observed behavior differences. This earlier
study used an ADHD composite code combin-
ing verbal, motor, and inattentive behaviors,
which did not permit examination of specific
behaviors that might underlie group differences.
In the current study, comparisons of more spe-
cific direct observation data revealed only one
statistically significant difference. That is, eth-
nic minority students exhibited higher levels of
off-task verbal behavior in both reading and
math. Results of the current study must consider
whether variations in teacher ratings reflect this
difference in observed verbal behavior.

Student Ethnicity

Prior investigations examining the influence

of student ethnicity on teacher ratings of ADHD
symptoms have produced mixed conclusions.
Sonuga-Barke et al. (1993) found teacher rat-
ings of ADHD symptoms in minority (“Asian”
students in the U.K.) students overestimated
true levels of behavior and provided evidence of
teacher bias. Similarly, Puig et al. (1999)
showed teacher ratings of overall problem be-
havior in African American students greatly
exaggerated observed levels of problem behav-
ior. In contrast, results of the current study did
not show signs of teacher bias manifested in

elevated ratings of ethnic minority students.
However, the current investigation differed
from Sonuga-Barke et al. in several important
ways, including the cultural context, ethnic
groups examined, and measures utilized. Evi-
dence of biased ratings from a U.K.-based study
may not generalize to United States culture.

Indeed, a recent U.S.-based study conducted

by Ramirez and Shapiro (2005) showed student
ethnicity (Hispanic or Caucasian) had no effect
on ADHD-IV ratings completed by Caucasian
teachers in an analog setting. Not only did the
current investigation provide further evidence
that ratings of Caucasian teachers (95.5% of
sample) are not biased against ethnic minority
(African American and Hispanic) students, it
also suggested teacher rating scale data may
more accurately reflect true behavioral levels of
these students as compared to their Caucasian
peers. Although only two studies comparing
teacher ratings of ADHD symptoms to direct
observations have been conducted in the U.S.,
neither has produced evidence of bias against
minority students in the form of inflated teacher
reports of ADHD symptoms.

Rater Effects

Numerous cross-cultural studies (e.g., Alban-

Metcalfe, Cheng-Lai, & Ma, 2002; Mueller et al.,
1995) have demonstrated the influence of rater
culture on perceptions of ADHD symptoms.
Ramirez and Shapiro (2005) provided evidence
for rater effects across ethnicities in the United
States. Specifically, their investigation showed
Hispanic teachers rated the behavior of His-
panic students more severely, suggesting they
hold stricter standards for behavior in students
of their own cultural background (Ramirez &
Shapiro, 2005). Unfortunately, the homogenous
sample of teachers in the current study does not
permit statistical analysis of effects of teacher
ethnicity. Nonetheless, the current results could
be explained by variation in teachers’ standards
for student behavior. One possible explanation
for results of the current study is that Caucasian
teachers hold students of their own ethnicity
to different behavioral standards. Significant
differences in correlations across student eth-
nicity suggest that observable ADHD symp-
toms of ethnic minority students are more
likely to be reflected in rating scale reports of
Caucasian teachers, whereas observable

430

HOSTERMAN, D

U

PAUL, AND JITENDRA

symptoms of Caucasian students may less
often manifest in teacher ratings.

This pattern is likely a mechanism of subtle

cultural perspectives. Past studies have demon-
strated that teachers are less tolerant of behav-
iors contrasting their own cultural expectations
(Gerber & Semmel, 1984; Lambert et al., 2001).
For example, it is possible that Caucasian teach-
ers are more sensitive and attuned to behaviors
of their ethnic minority students because the
movement styles, speech patterns, or manner-
isms of these students differ from their own.
This explanation is consistent with results of the
current study, in which off-task motor and off-
task verbal behaviors of ethnic minority stu-
dents explained significantly higher amounts of
variance in teacher’s ratings of hyperactive-
impulsive symptoms. Indeed, results of Puig
et al. (1999) showed teachers in the United
States may have a lower threshold of toler-
ance for behavior of African American stu-
dents. Ramirez and Shapiro (2005) describe
an adult threshold hypothesis in which teach-
ers from a certain culture will tolerate a range
of student behavior that remains within so-
cially established limits, but perceive any be-
haviors violating cultural standards as abnor-
mal. In other words, Caucasian teachers may
be habituated to off-task behaviors common
in children from their own culture. Behaviors
blending easily into a teacher’s familiar cul-
tural landscape may not feature prominently
during reflections on accumulated past expe-
riences with a target student. However, results
of this study do not suggest this heightened
sensitivity applies only to inappropriate class-
room behaviors, but show teachers are also
more likely to notice appropriate classroom
behaviors of ethnic minority students (e.g.,
active engagement).

Instrument Bias

Large scale studies of the CTRS and

ADHD-IV suggest these scales form similar
constructs and factors across ethnicities. How-
ever, this research also shows ratings of stu-
dents in certain ethnic groups are more likely to
fall in clinical ranges and be influenced by fac-
tors like SES, gender, and ethnicity (e.g., Reid
et al., 2000, 2001). Several investigators have
suggested separate norms may be needed for
gender and ethnicity, but all recognize the need

to rule out true behavioral difference through
direct observation (e.g., Epstein et al., 1998;
Reid et al., 2001). Within the current study, only
one significant difference between groups
emerged among the four ADHD indicators.
That is, students in the ethnic minority group
received significantly higher scores on the
ADHD-IV Hyperactive-Impulsive subscale in
both reading and math. However, direct obser-
vation data also showed that ethnic minority
students exhibited significantly higher levels of
off-task verbal behaviors, which are captured by
this scale. Thus, it appears this difference in
rating scale scores is likely rooted in true be-
havioral difference.

Although results of the current study do not

produce any evidence of inherent bias against
ethnic minority students in the CTRS and
ADHD-IV, the current data and sample are
clearly not sufficient to answer this question.
One study of the CTRS showed that when com-
pared to Caucasian students, African American
students received elevated scores, and Hispanic
students received lower scores (Langsdorf et al.,
1979). Analyses that combine these two groups,
as in the present study, cannot address the need
for separate norms for ethnicity. Further re-
search on the question of instrument bias and
need for ethnicity-specific norms is needed. Re-
search should also attend to the question of
whether ethnic differences impact a student’s
probability of meeting the clinical cutoff for
ADHD.

Disproportionate Referral and Placement

The hypothesis of bias in teacher ratings was

based on data showing that African American
and Hispanic students are referred for evalua-
tion and placed into special education at dispro-
portionately high rates (Coutinho & Oswald,
2000; Hosp & Reschly, 2003; Oswald,
Coutinho, Best, & Singh, 1999). Although con-
trary to the original hypothesis, results of the
current investigation are still consistent with
this referral pattern. This study does not sug-
gest teachers are likely overreferring ethnic
minority students by exaggerating their symp-
toms but may actually report behavior of
these students with greater accuracy. Instead,
results suggest disproportionate placement
may arise from underidentification and refer-
ral of Caucasian students relative to true lev-

431

TEACHER RATINGS OF ADHD IN ETHNIC MINORITY STUDENTS

els of observable symptoms. Similar results
were obtained in VanDerHeyden and Witt
(2005), who found teacher referrals of Cau-
casian students to the problem solving team
were less accurate than referrals of ethnic
minority students. Uneven distributions of re-
ferral across ethnicities may be influenced by
the tendency of Caucasian teachers to less
aptly identify problem behaviors in students
of their own culture. Because Caucasian
teachers are more likely to notice ADHD
symptoms in ethnic minority students, those
students may more often be referred and
placed into special education.

Limitations and Directions for

Future Research

The results and conclusions of the current

study must be considered in the context of sev-
eral limitations inherent in the design and sam-
ple. Results of this study have limited external
validity and may not generalize to schools in
other geographical areas, students of other eth-
nic backgrounds, or schools and classrooms
with different ethnic compositions (e.g., non-
Caucasian majorities). Nor can results be ex-
tended to teacher reports of different types of
student behaviors (e.g., internalizing symptoms)
or referrals (e.g., for academic concerns). It is
important that future investigations examine
whether similar patterns emerge in schools with
varying characteristics.

Construct validity must also be considered.

Although SES was treated as a covariate in this
study, other variables correlated with student
ethnicity (e.g., academic performance and im-
pact of a novel observer) may play a significant
role in influencing teacher ratings. Future stud-
ies should consider a broader range of student
characteristics and use statistical methods to
account for the influence of variables that vary
significantly across ethnic groups and may im-
pact teacher ratings. Similarly, ethnic composi-
tion of the school should be considered in future
studies because it can influence the impact of
SES (Hodgkinson, 1995).

A third and significant limitation of the cur-

rent study is the combination of African Amer-
ican and Hispanic students into a single ethnic
minority group. In this design, no conclusions
can be drawn for either ethnic group indepen-
dently, although it is very likely that teacher

perceptions of these two groups may differ sig-
nificantly. In fact, past research shows variation
in base rates of ADHD across Hispanic and
African American groups. The original research
question for this study focused on comparing
African American and Caucasian groups exclu-
sively. Because participants were drawn from
an existing sample of a larger study, combining
ethnic minority groups was necessary to obtain
sufficient power. This design also fails to ad-
dress the unique heterogeneity within each eth-
nic group. For example, within the Hispanic
population, influence of acculturation level, lan-
guage dominance, and country of origin may be
particularly salient. Cultural factors such as re-
ligion, parenting styles, and behavioral stan-
dards may also vary within groups and influence
teacher ratings. Future research should examine
similar questions in samples large enough to
support distinct groups of African American
and Hispanic students, and eventually examine
the unique dynamics within each ethnic group.
Thus, results of the current study can only sug-
gest presence of ethnic variation in accuracy of
teacher ratings but cannot inform specific
knowledge of between or within group differ-
ence and subtle cultural mechanisms.

A fourth limitation of this study was homo-

geneity in teacher ethnicity. Although the vast
majority of teachers across the nation are Cau-
casian, analyzing effects of other teacher per-
spectives is important to gaining a full under-
standing of the cultural dynamics involved in
teacher ratings of ADHD. Future investigations
should examine effects of varied combinations
of teacher and student ethnicities on accuracy of
rating scale data. The current study also did not
account for the ethnicities of individuals col-
lecting direct observation data. Although obser-
vation data are often assumed to be the gold
standard for measuring behavior, it is possible
that correlations between direct observation
data and teacher ratings were an artifact of
shared perspectives of teachers and raters from
common backgrounds. Future research should
systematically vary observer ethnicity and in-
clude IOA comparisons across raters of differ-
ent ethnicities to investigate the influence of this
variable and ensure objective observation data.

Future studies should be improved by closer

control and attention to several additional vari-
ables. Efforts should be made to minimize the
latency between direct observations and com-

432

HOSTERMAN, D

U

PAUL, AND JITENDRA

pletion of teacher ratings to ensure that interim
behaviors do not influence teacher perspectives.
Ideally, teachers would also remain blind to
student factors such as ADHD status and aca-
demic levels. Given the large number of com-
parisons completed in this study, the potential
influence of family wise (Type I) error rate
represents another limitation. Results would be
strengthened by greater statistical control for
family wise error rate. Finally, the effect of
teacher efficacy, a teacher’s beliefs that she can
influence a student’s learning ability and behav-
ior (Frey, 2002) warrants attention in future
investigations. Frey (2002) found teachers’ per-
ceptions of their classroom management and
discipline skills significantly predicted place-
ment referrals for behavior problems.

Implications for Practice

This investigation has several important im-

plications for practitioners in school and clinical
settings. Most importantly, results of this study
suggest school psychologists should interpret
ADHD rating scale data and teacher referrals
with caution. When interpreting these data, school
psychologists must consider the influence of eth-
nicity on teacher ratings. If results of the current
study are confirmed in later replications, practitio-
ners may lend more credence to Caucasian teach-
ers’ ratings of ethnic minority students. In con-
trast, knowing the same teacher’s rating of ADHD
symptoms in a Caucasian student may be less
sensitive to observable symptoms; the practitioner
may devote more time and consideration to direct
observation data.

These results also emphasize the importance

of using observational codes that compare be-
haviors of the target student to those of an average
peer. Use of local, within-classroom comparisons
will show whether the student’s behaviors are
truly unique among those of their classmates.
Such comparisons should also consider local ex-
pectations for student behavior and ethnic compo-
sitions of the classroom. Practitioners must also
attend to the possible influence of the rater’s eth-
nicity when reflecting on new referrals and inter-
preting rating scale data. These results emphasize
the importance of utilizing multimodal behavioral
assessment including varied approaches to mea-
surement, perspectives from multiple reporters
(including parents), and information on student
behavior across settings to ensure bias in any

single aspect of the assessment does not dominate
the final conclusion. In sum, knowledge of cul-
tural issues related to teachers’ reports of ADHD
symptoms is important to ensuring accurate diag-
noses of ADHD in different ethnic groups.

Conclusions

Contrary to the original hypothesis, results of

the current study suggested teacher’s ratings of
ADHD symptoms in ethnic minority students
are more accurate reflections of true behavioral
levels, based on direct observation data, than are
ratings of Caucasian students. Considering
Chang and Stanley’s (2003) definition of bias as
differing perceptions of “typical” and “atypical”
behavior across ethnic groups that results in
varying appraisal or actions based on the same
behavior sample, these findings suggest pres-
ence of “bias” in teacher ratings of ADHD
symptoms. However, this “bias” does not man-
ifest as the hypothesized exaggeration of symp-
toms in ethnic minority students, but rather as
possible underidentification of ADHD symp-
toms in Caucasian students.

Variation in teacher ratings of ADHD symp-

toms across student ethnicity is best explained
by an interaction between rater and student eth-
nicities. The question of teacher bias and varied
standards for student behavior will only be re-
solved through more studies of this nature. Fur-
ther research is needed to determine how accu-
racy of rating scale data is impacted by varied
combinations of teacher and student ethnicities.

References

Alban-Metcalfe, J., Cheng-Lai, A., & Ma, T. (2002).

Teacher and student teacher ratings of Attention-
Deficit/Hyperactivity Disorder in three cultural
settings. International Journal of Disability, De-
velopment and Education, 49, 281–298.

American Psychiatric Association. (2000). Diagnos-

tic and statistical manual of mental disorder. (4th
ed., text revision) Washington, DC: Author.

Anastopoulos, A. D., & Shelton, T. L. (2001). As-

sessing attention-deficit/hyperactivity disorder.
New York, NY: Kluwer Academic/Plenum Press.

Arnold, L. E., Elliott, M., Sachs, L., Bird, H., Krae-

mer, H. C., Wells, K. C. et al. (2003). Effects of
ethnicity on treatment attendance, stimulant re-
sponse/dose, and 14-month outcome in ADHD.
Journal of Consulting and Clinical Psychology,
71, 713–727.

433

TEACHER RATINGS OF ADHD IN ETHNIC MINORITY STUDENTS

Artiles, A. J., & Trent, S. C. (1994). Overrepresen-

tation of minority students in special education: A
continuing debate. Journal of Special Educa-
tion, 27, 410 – 437.

Bahr, M. W., Fuchs, D., Strecker, P. M., & Fuchs, L. S.

(1991). Are teachers’ perceptions of difficult-to-teach
students racially biased? School Psychology Re-
view, 20, 599 – 608.

Barkley, R. A. (2006). Attention deficit hyperactivity

disorder: A handbook for diagnosis and treatment
(3rd ed.). New York, NY: Guilford Press.

Busse, R. T., & Beaver, B. R. (2000). Informant

report: Parent and teacher interviews. In E. S.
Shapiro & T. R. Kratochwill (Eds.), Conducting
school-based assessments of child and adolescent
behavior (pp. 235–273). New York, NY: Guilford
Press.

Chang, D. F., & Stanley, S. (2003). The effects of

race and problem type on teacher’s assessments of
student behavior. Journal of Consulting and Clin-
ical Psychology, 71, 235–242.

Cohen, J. (1988). Statistical power analysis for the

behavioral sciences (2nd ed.). Hillsdale, NJ: Erl-
baum, Publishers.

Cohen, J., & Cohen, P. (1983). Applied multiple

regression/correlation analysis for the behavioral
sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Conners, C. K. (1973). Rating scales for use in drug

studies with children. Psychopharmacology Bulle-
tin, 9, 24 –29.

Conners, C. K. (1997). Conners’ Rating Scales-

Revised. North Tonawanda, NY: Multi Health
Systems.

Coutinho, M. J., & Oswald, D. P. (2000). Dispropor-

tionate representation in special education: A syn-
thesis and recommendations. Journal of Child and
Family Studies, 9, 135–156.

Cullinan, D., & Kauffman, J. M. (2005). Do race of

student and race of teacher influence ratings of
emotional and behavioral problem characteristics
of students with emotional disturbance? Behav-
ioral Disorders, 30, 393– 402.

DiPerna, J., & Elliot, S. (2000). Academic compe-

tence evaluation scales. San Antonio, TX: Har-
court Assessment.

Donovan, M. S., & Cross, C. T. (Eds.). (2002). Mi-

nority students in special and gifted education.
Washington DC: National Academic Press.

DuPaul, G. J., Jitendra, A. K., Volpe, R. J., Tresco,

K. E., Lutz, J. G., Vile Junod, R. E., et al. (2006).
Consultation-based academic interventions for
children with ADHD: Effects on reading and
mathematics achievement. Journal of Abnormal
Child Psychology, 34, 633– 646.

DuPaul, G. J., Power, T. J., Anastopoulos, A., &

Reid, R. (1998). ADHD Rating Scale IV Manual.
New York, NY: Guilford Press.

DuPaul, G. J., & Stoner, G. (2003). ADHD in the

schools: Assessment and intervention strategies
(2nd ed.). New York, NY: Guilford Press.

Epstein, J. N., March, J. S., Conners, C. K., & Jackson,

D. L. (1998). Racial differences on the Conners Teacher
Rating Scale. Journal of Abnormal Child Psychol-
ogy, 26, 109–118.

Epstein, J. N., Willoughby, M., Valencia, E. Y.,

Tonev, S. T., Abikoff, H. B., Arnold, L. E., et al.
(2005). The role of children’s ethnicity in the re-
lationship between teacher ratings of Attention-
Deficit/Hyperactivity Disorder and observed class-
room behavior. Journal of Consulting and Clinical
Psychology, 73, 424 – 434.

Frey, A. (2002). Predictors of placement recommen-

dations for children with behavioral or emotional
disorders. Behavioral Disorders, 27, 126 –136.

Gerber, M. M., & Semmel, M. I. (1984). Teacher as

imperfect test: Reconceptualizing the referral pro-
cess. Educational Psychologist, 19, 137–138.

Hodgkinson, H. L. (1995). What should we call peo-

ple? Race, class, and the census for 2000. Phi
Delta Kappan, 77, 173–179.

Hollingshead, A. B. (1976). Four-factor index of

social status. New Haven, CT: Yale University
Press.

Hosp, J. L., & Reschly, D. J. (2003). Referral rates

for intervention or assessment: A meta analysis of
racial differences. The Journal of Special Educa-
tion, 37, 67– 80.

Hosp, J. L., & Reschly, D. J. (2004). Disproportion-

ate representation of minority students in special
education: Academic, demographic, and economic
predictors. Exceptional Children, 70, 185–199.

Lambert, M. C., Puig, M., Lyubansky, M., Rowan,

G. T., & Winfrey, T. (2001). Adult perspectives on
behavior and emotional problems in African
American children. Journal of Black Psychol-
ogy, 27, 64 – 85.

Langsdorf, R., Anderson, R. P., Waechter, D.,

Madrigal, J. F., & Juarez, L. J. (1979). Ethnicity,
social class, and perception of hyperactivity.
Psychology in the Schools, 16, 293–298.

Mann, E. M., Ikeda, Y., Mueller, C. W., Takashashi,

A., Tao, K. T., Humris, E., et al. (1992). Cross-
cultural differences in rating hyperactive-disrup-
tive behaviors in children. American Journal of
Psychiatry, 149, 1539 –1542.

Mueller, C. W., Mann, E. M., Thanapum, S., Humris,

E., Ikeda, Y., Takahashi, A., et al. (1995). Teachers’
ratings of disruptive behavior in five countries. Jour-
nal of Clinical Child Psychology, 24, 434 – 442.

National Center for Educational Statistics (2001).

Digest of Educational Statistics, 2001. [Electronic
Version]. Washington, DC: U.S. Department of
Education.

Neal, L. V. I., McCray, A. D., Webb-Johnson, G., &

Bridgest, S. T. (2003). The effects of African

434

HOSTERMAN, D

U

PAUL, AND JITENDRA

American movement styles on teachers’ percep-
tions and reactions. The Journal of Special Edu-
cation, 37, 49 –57.

Oswald, D. P., Coutinho, M. J., Best, A. M., & Singh,

N. N. (1999). Ethnic representation in special ed-
ucation: The influence of school-related economic
and demographic variables. Journal of Special Ed-
ucation, 32, 194 –206.

Pelham, W. E. Jr., Milich, R., Murphy, D., & Mur-

phy, H. A. (1989). Normative data on the IOWA
Conners Teacher Rating Scale. Journal of Clinical
Child Psychology, 18, 259 –262.

Prieto, A. G., & Zucker, S. H. (1981). Teacher per-

ception of race as a factor in the placement of
behaviorally disordered children. Behavioral Dis-
orders, 7, 34 –38.

Puig, M., Lambert, M. C., Rowan, G. T., Winfrey, T.,

Lyubansky, M., Hannah, S. D., et al. (1999). Be-
havioral and emotional problems among Jamaican
and African American children, ages 6 to 11:
Teacher reports versus direct observations. Jour-
nal of Emotional and Behavioral Disorders, 7,
240 –250.

Ramirez, R. D., & Shapiro, E. S. (2005). Effects of

student ethnicity on judgments of ADHD symp-
toms among Hispanic and White Teachers. School
Psychology Quarterly, 20, 368 –287.

Reid, R., Casat, C. D., Norton, H. J., Anastropoulos,

A. D., & Temple, E. P. (2001). Using behavior
rating scales for ADHD across ethnic groups: The
IOWA Conners. Journal of Emotional and Behav-
ioral Disorders, 9, 210 –218.

Reid, R., DuPaul, G. J., Power, T. J., Anastopoulos,

A. D., Rogers-Adkinson, D., Noll, M-B., et al.
(1998). Assessing culturally different students for
attention deficit hyperactivity disorder using be-
havior rating scales. Journal of Abnormal Child
Psychology, 26, 187–198.

Reid, R., Maag, J. W., Vasa, S. F., & Wright, G.

(1994). Who are the children with ADHD: A
school-based survey. Journal of Special Educa-
tion, 28, 117–137.

Reid, R., Riccio, C. A., Kessler, R. H., DuPaul, G. J.,

Power, T. J., Anastopoulos, A. D., et al. (2000).
Gender and ethnic difference in ADHD as assessed
by behavior ratings. Journal of Emotional and
Behavioral Disorders, 8, 38 – 48.

Shaffer, D., Fisher, P., & Lucas, M. K. (1998). NIMH

Diagnostic Interview Schedule for Children
(DISC-IV). New York, NY: New York State Psy-
chiatric Institute.

Shapiro, E. S. (2003). Behavioral Observation of

Students in Schools-BOSS (Computer Software).
San Antonio, TX: Psychological Corporation.

Shapiro, E. S., & Kratochwill, T. R. (2000). Intro-

duction: Conducting a multidimensional behav-
ioral assessment. In E. S. Shapiro & T. R. Kratoch-
will (Eds.), Conducting school based assessments
of child and adolescent behavior (pp. 1–20). New
York, NY: Guilford Press.

Sonuga-Barke, E. J., Minocha, K., Taylor, E. A., &

Sandberg, S. (1993). Inter-ethnic bias in teachers’
ratings of childhood hyperactivity. British Journal
of Developmental Psychology, 11, 187–200.

Stevens, G. (1980). Bias in attributions of positive

and negative behavior in children by school psy-
chologists, parent, and teachers. Perceptual and
Motor Skills, 50, 1283–1290.

Swanson, J. M. (1992). School-based assessments

and interventions for ADD students. Irvine, CA:
K. C. Publishing.

United States Department of Education, Office of

Educational Research and Improvement. (1998).
The schools and staffing survey (SASS) and
teacher follow-up survey (TFS) CD ROM: Elec-
tronic codebook and public-use data for three
cycles of SASS and TRS. Washington, DC: Na-
tional Center for Educational Statistics. (NCES:
98 –312).

University of Amsterdam Faculty of Humanities (n. d.)

Two correlation coefficients (online calculator). Re-
trieved July 12, 2006, from http://www.fon.hum.uva.nl/
Service/Statistics/Two_Correlations.html

VanDerHeyden, A. M., & Witt, J. C. (2005). Quan-

tifying context in assessment: Capturing the effect
of base rates on teacher referral and problem-
solving model of identification. School Psychology
Review, 34, 161–183.

Weisz, J. R., Chaiyasit, W., Wiss, B., Eastman, K. L.,

& Jackson, E. W. (1995). A multimethod study of
problem behavior among Thai and American chil-
dren in school: Teacher reports versus direct ob-
servations. Child Development, 66, 402– 415.

Woodcock, R. W., McGrew, K. S., & Mather, N.

(2001). Woodcock-Johnson III Tests of Achieve-
ment. Itasca, IL: Riverside Publishing.

Ysseldyke, J. E., Vanderwood, M. L., & Shriner, J.

(1997). Changes over the past decade in special
education referral to placement probability: An
incredibly reliable practice. Diagnostique, 23,
193–203.

435

TEACHER RATINGS OF ADHD IN ETHNIC MINORITY STUDENTS