Disentangling teacher and lesson variance and their relations to supportive learning environments: A machine learning assessment of teacher nonverbal behavior.

Autor/inn/en	Lazarides, Rebecca; Frenkel, Jonas; Daumiller, Martin; Dresel, Markus; Petkovic, Uros; Hellwich, Olaf; Göllner, Richard
Titel	Disentangling teacher and lesson variance and their relations to supportive learning environments: A machine learning assessment of teacher nonverbal behavior. Gefälligkeitsübersetzung: Entflechtung von Lehrer- und Unterrichtsunterschieden und deren Zusammenhang mit unterstützenden Lernumgebungen: Eine maschinelle Bewertung des nonverbalen Verhaltens von Lehrern.
Quelle	In: The journal of educational psychology, 118 (2026) 6, S. 976-999Infoseite zur Zeitschrift PDF als Volltext Link als defekt meldenVerfügbarkeit
Sprache	englisch
Dokumenttyp	online; Zeitschriftenaufsatz
ISSN	1939-2176
DOI	10.1037/edu0000999
Schlagwörter	Nonverbale Kommunikation; Verhalten; Lernklima; Lehrer; Lernumgebung; Unterricht; Unterstützung
Abstract	Teachers? nonverbal behaviors are critical for motivating students and fostering learning. However, there is a lack of research considering the situatedness of nonverbal teaching behaviors. Further, the question remains as to how nonverbal behaviors interrelate with dimensions of supportive teaching. We utilized a machine learning approach to assess teachers? nonverbal immediacy (NVI) from classroom video data and investigated its variability across learning situations, lessons, and teachers. Additionally, we examined the relations between teachers? NVI and observer-rated social?emotional support at the level of the teacher, the lesson, and the learning situation. We also investigated relations between NVI and student?teacher relationships and autonomy support focused on the lessons of the specific teaching unit as reported by the teacher and the students after the lessons. Data stemmed from the German Teaching and Learning International Survey (TALIS) video study (Klieme et al., 2023), which was implemented as a national sub-study of the international TALIS video study. The sample of the German TALIS video study included 50 teachers (46% women; M age = 43 years) and their 1,140 students (53% girls; M age = 15 years). Machine learning-assessed NVI showed 22.5% between-teacher and 5.6% between-lesson variance, with ~71.7% at the situational level. Externally observed socioemotional support showed higher variance between teachers (39.9%) and lessons (25.5%), but lower variance at the situational level (~34.5%). Results from three-level modeling revealed that teachers? NVI differed substantially between teachers and lessons. Teachers? NVI was significantly positively associated with observer-rated socioemotional support across all levels of analysis. At the teacher level, teachers? NVI was significantly and positively related to autonomy support. Our findings indicate that nonverbal behaviors of the teacher assessed using computational modeling are substantially related to a supportive classroom climate. (ZPID).
Erfasst von	Leibniz-Institut für Psychologie, Trier
Update	2026/3