stanford-nlpxed/classroom_management_data

---
license: mit
task_categories:
- text-classification
language:
- en
size_categories:
- 1K<n<10K
---

# Dataset Card for Classroom Management

## Dataset Summary
Three annotated datasets of teachers' classroom management language. The annotations identify utterances as 1) classroom management, 2) behavior management, and 3) specific behavior management talk moves such as command, reprimands, threats, and exclusionary consequences.
This repository also contains two notebooks for training RoBERTa models using this data, `train_binary.ipynb` and `train_multiclass.ipynb`.

## Supported Tasks
Classification of teachers' classroom management talk moves in utterances from elementary math classroom transcripts

## Languages
English

## Dataset Description
### File 1: classroom_management.csv
- Number of instances: 10354
- Features/Columns: 2

#### Data Fields
- `text` — 7 utterances, composed of a target utterance marked by [TARGET][/TARGET] and the 3 utterances preceding and following. Utterances include speaker markings, but the target utterance is always the teacher.
- `labels` — 1 if the utterance is an instance of classroom management language, 0 otherwise

### File 2: behavior_management.csv
- Number of instances: 8057
- Features/Columns: 2

#### Data Fields
- `text` — 7 utterances, composed of a target utterance marked by [TARGET][/TARGET] and the 3 utterances preceding and following. Utterances include speaker markings, but the target utterance is always the teacher.
- `labels` — 1 if the utterance is an instance of behavior management language, 0 otherwise

### File 3: talkmoves.csv
- Number of instances: 5720
- Features/Columns: 7

#### Data Fields
- `text` — A single teacher utterance, pre-determined as being an instance of behavior management language.
- `vs` — 1 if the utterance is an instance of verbal sanctioning, 0 otherwise
- `vs_submove` — 1 if the utterance is an instance of a simple desist, 2 if the utterance is an instance of a command, 3 if the utterance is an instance of a reprimand, 4 if the utterance is an instance of a threat
- `ms` — 1 if the utterance is an instance of material sanctioning, 0 otherwise
- `ms_excl` — 1 if the utterance is an instance of the kind of material sanction that involves exclusionary discipline, 0 otherwise
- `ms_excl_iso` — 1 if the utterance is an instance of exclusionary discipline that involves isolating the student outside the classroom, 0 otherwise
- `ms_excl_ch` — 1 if the utterance is an instance of exclusionary discipline that involves calling home, 0 otherwise

## Data Collection
Transcript data from the NCTE dataset [1] transcribed from the NCTE main study [2].
- [1] Demszky, D., & Hill, H. (2023). The NCTE Transcripts: A Dataset of Elementary Math Classroom Transcripts. In 18th Workshop on Innovative Use of NLP for Building Educational Applications.
- [2] Kane, Thomas, Hill, Heather, and Staiger, Douglas. National Center for Teacher Effectiveness Main Study. Inter-university Consortium for Political and Social Research [distributor], 2022-06-16. https://doi.org/10.3886/ICPSR36095.v4

## Annotation
Annotators are 6 experienced elementary math teachers trained on this task. Fleiss' Kappa scores range from 0.719 to 0.964.

## Ethical Considerations
Not intended for evaluation of teaching quality. What is appropriate in a given classroom is highly contextual and relational in a way that these annotations to not capture.

## Citation
If you use this dataset, please cite:

Tan, Mei, and Dorottya Demszky. (2025). Do As I Say: What Teachers’ Language Reveals About Classroom Management Practices. (EdWorkingPaper: 23-844). Retrieved from Annenberg Institute at Brown University: https://doi.org/10.26300/9yj6-jn52