Data Dictionary: student_engagement_ordinal.csv

Ordinal outcome dataset for optional Bayesian cumulative models.

Use this dataset for ordinal-outcome teaching. It supports the shift from coefficient reporting to predicted category probabilities.

Research Scenario

Students report engagement on a five-point Likert scale. A researcher asks whether an intervention is associated with higher engagement after accounting for baseline motivation and teacher feedback.

Variables

Variable Type Description
respondent_id string Unique respondent identifier.
intervention binary 1 if student received the intervention, 0 otherwise.
baseline_motivation integer Baseline motivation rating from 1 to 5.
teacher_feedback integer Perceived teacher feedback rating from 1 to 5.
engagement ordinal integer Engagement response from 1 to 5.

Main Questions

  • How can Bayesian models handle ordinal outcomes without treating Likert responses as continuous by default?
  • How can category probabilities be interpreted?
  • How can posterior predictions communicate expected shifts in response categories?