1. • Approximately 1 in 20 children in America have a diagnosable mental
health problem (Kovacs et al. 2014). However, many mental health
problems, including attention deficit hyperactivity disorder (ADHD) go
undiagnosed and untreated (Merikangas et al., 2011).
• Recently, attention has been devoted to the use of school-based
screening as a mechanism to identify social, emotional, and/or
behavioral problems and those in need of services (Dowdy et al., 2014),
yet much of this research has focused on elementary school students.
• We sought to examine the psychometric properties of two screening
tools that include assessment of social, emotional and behavioral
problems when completed by high school teachers. It is important to
understand this information prior to using such tools to identify high
school students in need.
Psychometric Properties of Two Teacher-Rated
Screening Tools for Identifying ADHD in High
School Youth
Introduction
Annalise M. McMahan1, Gina M. Sacchetti1, Madeleine E. Schwartz1, Steven W. Evans1, Eloise Kaizar2
and Julie Sarno Owens1
1Ohio University; Ohio State University2
Method
Results
Discussion
Participants
• We obtained teacher ratings on 1,626 high school students, aged 14-19
(94% response rate; 53.6% male) from one school district in a small city in
Ohio. The school district is characterized as Caucasian (95%), with 48%
eligible for free/reduced lunch, and 15% eligible for special education.
Measures
• Teachers completed (a) the Strengths and Difficulties questionnaire (SDQ;
Goodman, 2001), which produces a total score and five subscale scores:
emotion, conduct, hyperactivity-inattention, peer relations, and prosocial;
and (b) the Behavioral and Emotional Screening System (BESS; Kamphaus
& Reynolds, 2007) which produces a total T-score based on items assessing
adjustment, externalizing, internalizing, and school problems. Both
measures are designed for use with high school students.
Procedure
• To select one teacher to complete the on-line REDCap survey for each
student, we used the following guidelines: (a) Each teacher was asked to
complete surveys for no more than 30 students; (b) Teachers of study halls,
physical education, music, or on-line courses were not asked to complete
surveys; and (c) the student’s first period teacher was invited first. If that
teacher declined or the student was not on campus first period, the teacher
of the second period class was invited, and so forth until there was a teacher
match for each student that meets the above rules.
This presentation was supported by the Disability Research and Dissemination Center (DRDC) through its Grant Number 5U01DD001007-03, FAIN No. U01DD001007 from the Centers for Disease Control and Prevention (CDC). Its contents are
solely the responsibility of the authors and do not necessarily represent the official views of the DRDC or the CDC
• These data suggest that both measures have acceptable internal
consistency and that their scores are highly related.
• However, the measures identify different rates of students (i.e., 21.7%
vs. 27.7%) and the students they identify may be meaningfully different.
• Namely, because the SDQ has more items devoted to social functioning,
the SDQ may be better at detecting youth with social impairment than
academic impairment. Conversely, because the BESS has more items
devoted to school functioning, its scores may be better at detecting
youth with academic impairment.
• Future research should examine the extent to which each of these
measures predicts important academic, behavioral and social outcomes.
• The internal consistency estimates for the BESS total was α = .95, for
the SDQ total was α = .88; the SDQ subscales were .89 (prosocial), .85
(inattention/hyperactivity), .82 (conduct), .80 (emotion), and .70 (peer).
• The correlation between the BESS and the SDQ total was .88, with
correlations with the SDQ subscales ranging from .48 (peer) to .80
(hyperactivity-inattention).
• Using the recommended cut-off scores for risk identification (i.e., a
BESS T-score of 61 or higher; SDQ total score of 12 or higher, which
represents a t-score of 58), the BESS identified 27.7% of students as at
risk and the SDQ identified 21.7% as at risk.
• A chi square test revealed that BESS risk status and SDQ risk status are
associated, X2 (1, 1626) = 754.574, p < .001. Yet, agreement is only
moderate (kappa= .67). Namely, 148 students (9.1%) were identified as
at risk by the BESS but not the SDQ, and 51 students (3.1%) were
identified as at risk by the SDQ but not the BESS (see Table 1).
• Interestingly, those identified only by the SDQ had significantly greater
impairment in peer relations t(110) = 3.15 (p<.01), but significantly less
impairment in the classroom t(110) = -5.69 (p<.01), as measured by the
SDQ impairment items (see Table 2), relative to students identified only
by the BESS.
Typical BESS At-Risk BESS
Typical SDQ 1125 (69.2%) 148 (9.1%)
At-Risk SDQ 51 (3.1%) 302 (21.7%)
Table 1. Student Risk Status as a Function of Rating Scale
SDQ Peer
Impairment
SDQ Classroom
Impairment
At-Risk on SDQ Only 1.26 (.73) .95 (.62)
At-Risk on BESS Only 0.74 (.64) 1.98 (.74)
Table 2. Impairment Ratings by Students At Risk by SDQ or BESS only
Note. Scores range from 0 (not at all) to 3 (a great deal)