Application of Bayesian LGCM to study students with SEBD

Inge Zweers-Schrooten, Rens Van de Schoot, Nouchka Tick, Sarah Depaoli, James P. Clifton, Bram Orobio de Castro, Jan O. Bijstra

Research output: Contribution to conferencePaperAcademicpeer-review

Abstract

Students with emotional and behavioural disorders (EBD) comprise a small, exceptional student population. We compared the development of student-teacher conflict among (a) students with EBD included in regular classrooms (included; n = 39), (b) students with EBD in exclusive schools for special education (excluded; n = 15) and (c) typically developing controls (controls; n = 1321).
We collected multi-informant longitudinal survey data on student-teacher conflict and on various aspects that may predict development in student-teacher relationships. We set up a three phase Bayesian latent growth model (see Figure 1) with informative priors to examine students’ development in student-teacher conflict. We examined differences in intercepts and slopes between the three subgroups (phase 2) and we focused on included and excluded students (phase 3), to see whether we could find predictors of growth in student-teacher
conflict.
Excluded students reported more student-teacher conflict than controls, and excluded students reported decreasing conflict over time, while conflict remained stable over time among included students and controls. These results were stable from a sensitivity analysis. For both included and excluded students, student-teacher conflict intercepts were only predicted by sex and prior student-teacher conflict. The results of the model with these small subgroups were unstable from a sensitivity analysis.
This study shows that with Bayesian latent growth modelling, we can obtain stable results for small, exceptional populations like students with EBD. That is, excluded students had more conflictual relationships with teachers than controls, but relationships of the former subgroup improved over time while those of the latter subgroup remained stable over time. However, to run more complex models with these exceptional small groups, we will need more prior information to obtain stable results.
Original languageEnglish
Publication statusPublished - 1 Sep 2017
Externally publishedYes

Keywords

  • social-emotional/behavioral difficulties
  • student-teacher relationship
  • peer acceptance
  • self-esteem
  • Bayesian Statistics

Fingerprint Dive into the research topics of 'Application of Bayesian LGCM to study students with SEBD'. Together they form a unique fingerprint.

Cite this