Poster #119 - Teacher Evaluations of Student Behavior in High Poverty Urban Schools: An Investigation of Measurement Invariance

• In Event: 3-053 - Poster Session 10
  In Poster Session: PS 10 Section - Methods, History, Theory

Integrative Statement

A significant challenge faced in schools is promoting positive educational and mental health outcomes for students characterised as having emotional behavioural difficulties (Roeser, 2001). Recent research demonstrates how evaluative tools of student behaviour can enable teachers and educational professionals to gain a broad view of behaviour, remediate behaviour problems and promote a classroom environment in which students may thrive (Hughes et al., 2015; Janus & Offord, 2007; Kamphaus et al., 2007; Mahone & Hoffman, 2007). With this in mind, the current study used measurement invariance testing procedures to examine the psychometric properties of one such measure of student behaviour, the Behavior Assessment System for Children, Second Edition- Teacher Rating Scales (BASC-2 TRS; Reynolds & Kamphaus, 2004).

This study investigated teacher evaluations of student behaviour in an urban, high poverty, ethnic minority sample (n = 484; Xyears =10.14, SDyears = 1.13). Pre-existing data was used from a project exploring the malleability of executive functions. Prior to analysing measurement invariance, independent samples t-tests of each subscale were run to explore whether the sample was comparable to a representative random sample within the United States, or if it was distinct as a targeted sample from a high-poverty area (see Table 1). Measurement invariance was then evaluated for each composite scale of the BASC-2 TRS when completed by the same teacher at the beginning and end of an academic year. Finally, the BASC- 2 TRS composites were evaluated for measurement invariance as completed by two different teachers one year apart.

All subscale means for the study sample were significantly different from the normed sample. The majority of these differences are modest, as evidenced by small effect sizes, however, analyses of “Functional Communication” (d = .80) and “Adaptability” (d = .93) yielded large effect sizes, suggesting that differences in teacher ratings on these composite variables are highly perceptible. In addition, measurement invariance was established for three of the five composite scales within the BASC- 2 TRS (Externalising Problems, School Problems, and Adaptive Skills). This invariance remained consistent longitudinally and across teacher rating groups. Measurement invariance was not demonstrated for the Internalising Problems composite scale nor for the Behavioural Symptoms Index composite scale. Research questions and hypotheses in the current study were based on the BASC-2 TRS as a whole, and we did not foresee factional invariance among the composite scales. As such, further exploration of variance within the Internalising Problems and the Behavioural Symptoms Index composite scales is warranted. Future research using exploratory factor analyses to determine if the subscales and items for these composite scales should, in fact, be included may also be useful. Additionally, researchers should consider replication studies exploring whether the current results emerge with different age groups and across gender. Despite limitations, this project demonstrates the complexities of measurement invariance research and highlights the importance of using measurement invariance to establish consistency in behavioural assessments.

Authors

• Anne Katharine Zonneveld, University of Cambridge
  Presenting Author

• Zewelanji Serpell, Virginia Commonwealth University
  Non-Presenting Author

• Teresa Parr, Virginia Commonwealth University
  Non-Presenting Author

• Michelle Ellefson, University of Cambridge
  Non-Presenting Author