Intra-individual variability of disruptive behavior in elementary school: a Bayesian location-scale analysis of daily direct behavior ratings

Introduction Disruptive behavior in elementary school classrooms fluctuates from day to day. We ask whether the size of these fluctuations differs systematically among students and what such differences reflect. Methods We analyzed baseline data from three single-case intervention studies: 739 daily direct behavior ratings (DBR) of disruptive behavior by class teachers for 37 elementary school students (ages 6 to 9) with externalizing behavior problems in 13 classes. Bayesian mixed-effects location-scale models estimated each student's level and within-person variability jointly. Robustness was examined with an ordinal likelihood, autoregressive residuals, alternative priors, other inclusion thresholds, and classroom-level random effects. Results Within-person variability differed substantially among the rated series, beyond sampling error: on the 0 to 10 scale, the most variable student fluctuated about 9.7 times more strongly than the most stable one (95% credible interval 6.2 to 15.4). Models with student-specific variability outperformed common-variance models in observation-level and student-level cross-validation and under all robustness checks. The central qualification concerns attribution: with a classroom level in the model, most of the variance of the log-scale effects was attributed to the classroom, which confounds rating teacher, classroom context, and student selection (wide uncertainty given 13 classes). Students rated near the floor or ceiling also showed compressed variability. Variability was unrelated to the mean level in the interior of the scale; its relation to teacher-rated severity of oppositional and disruptive behavior could not be determined with 24 students, as the credible interval permits meaningful effects in either direction. Its consistency across the two baseline halves was positive but imprecise. Once level and trend were accounted for, ratings showed only weak day-to-day dependence (lag-1 autocorrelations of .02 and .04; model-based autoregressive parameters of .20 and .28), whereas pooling across students inflated it severalfold. Discussion Systematic differences in within-person variability are a property of daily rating series, distinct from their level, but ratings from one teacher per class cannot attribute them to the student rather than to the class or the rater. Single ratings carry little information, and baseline phases in single-case research need enough measurement points to estimate a student's level precisely.

Authors

Institutions

Publication Details

Journal
Frontiers in Education
Published
2026-11-13
DOI
https://doi.org/10.3389/feduc.2026.1973541
Primary Topic
Behavioral and Psychological Studies
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Intra-individual variability of disruptive behavior in elementary school: a Bayesian location-scale analysis of daily direct behavior ratings

Tobias Hagen, Jannik Nitz, Thomas Hennemann
Frontiers in Education
Behavioral and Psychological Studies
article

Intra-individual variability of disruptive behavior in elementary school: a Bayesian location-scale analysis of daily direct behavior ratings

Tobias Hagen, Jannik Nitz, Thomas Hennemann
article en

Abstract

Introduction Disruptive behavior in elementary school classrooms fluctuates from day to day. We ask whether the size of these fluctuations differs systematically among students and what such differences reflect. Methods We analyzed baseline data from three single-case intervention studies: 739 daily direct behavior ratings (DBR) of disruptive behavior by class teachers for 37 elementary school students (ages 6 to 9) with externalizing behavior problems in 13 classes. Bayesian mixed-effects location-scale models estimated each student's level and within-person variability jointly. Robustness was examined with an ordinal likelihood, autoregressive residuals, alternative priors, other inclusion thresholds, and classroom-level random effects. Results Within-person variability differed substantially among the rated series, beyond sampling error: on the 0 to 10 scale, the most variable student fluctuated about 9.7 times more strongly than the most stable one (95% credible interval 6.2 to 15.4). Models with student-specific variability outperformed common-variance models in observation-level and student-level cross-validation and under all robustness checks. The central qualification concerns attribution: with a classroom level in the model, most of the variance of the log-scale effects was attributed to the classroom, which confounds rating teacher, classroom context, and student selection (wide uncertainty given 13 classes). Students rated near the floor or ceiling also showed compressed variability. Variability was unrelated to the mean level in the interior of the scale; its relation to teacher-rated severity of oppositional and disruptive behavior could not be determined with 24 students, as the credible interval permits meaningful effects in either direction. Its consistency across the two baseline halves was positive but imprecise. Once level and trend were accounted for, ratings showed only weak day-to-day dependence (lag-1 autocorrelations of .02 and .04; model-based autoregressive parameters of .20 and .28), whereas pooling across students inflated it severalfold. Discussion Systematic differences in within-person variability are a property of daily rating series, distinct from their level, but ratings from one teacher per class cannot attribute them to the student rather than to the class or the rater. Single ratings carry little information, and baseline phases in single-case research need enough measurement points to estimate a student's level precisely.

Frontiers in EducationVol. 11
Carl von Ossietzky Universität Oldenburg (DE), University of Cologne (DE)
Quality education
Behavioral and Psychological Studies
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon