Skip to main content Skip to search
Predictors of Rater Bias in the Assessment of Social-Emotional Competence
International Journal of Emotional Education
Short Title: International Journal of Emotional Education
Format: Journal Article
Publication Date: 2016/11/01/
Pages: 25 - 44
Sources ID: 90371
Notes: Accession Number: EJ1121269; Acquisition Information: Centre for Resilience and Socio-Emotional Health. Old Humanities Building (OH) Room 241, University of Malta, MSD 2080, Malta. Tel: +356-2340-3014; Web site:; Language: English; Education Level: Elementary Education; Reference Count: 50; Journal Code: APR2018; Level of Availability: Available online; Publication Type: Academic Journal; Publication Type: Report; Entry Date: 2016
Visibility: Public (group default)
Abstract: (Show)
"The Devereux Student Strengths Assessment Mini" (DESSA-Mini) (LeBuffe, Shapiro, & Naglieri, 2014) efficiently monitors the growth of Social-Emotional Competence (SEC) in the routine implementation of Social Emotional Learning programs. The DESSA-Mini is used to assess approximately half a million children around the world. Since behavior rating scales can have "rater bias", this paper examines rater characteristics that contribute to DESSA-Mini ratings. Rater characteristics and DESSA-Mini ratings were collected from elementary school classroom teachers (n = 72) implementing TOOLBOX in a racially/ethnically diverse California school district. Teachers rated 1,676 students, who scored similarly to a national reference group. Multilevel modeling analysis showed that only 16% of variance in DESSA-mini ratings was attributable to raters. Relationships between teacher characteristics and ratings were estimated to examine rater variance. Collectively, four characteristics of teachers (perceived barriers to student learning, sense of their "typical" student's level of SEC, anticipation of SEL program implementation challenges, and intentions to fully implement a newly adopted SEL program) accounted for bias in teacher-generated DESSA scores, leaving only 10% of the variance unexplained. Identified sources of "rater bias" can be controlled for in research and addressed through thoughtful program selection, training, and implementation.