AI Self-Efficacy Predictor
Set your ChatGPT literacy across five dimensions and walk the study's decision tree to predict AI self-efficacy — then see what drives the model and how to improve.
What the study found
Non-linearAcross 393 Romanian educators, a Decision Tree Regression modelled AI self-efficacy from five ChatGPT literacy dimensions. Technical proficiency — the ability to write and refine effective prompts — was by far the strongest predictor, and the very first split in the tree. The model explained about 53% of the variance in self-efficacy.
Model performance
Trained on 315 educators (80%), validated on 78 (20%). Predictors were standardized (z-scores), so the tree's split points are expressed in standard-deviation units. Explore the modules below.
Your ChatGPT literacy
1–5 LikertRate yourself on each dimension of the ChatGPT Literacy Scale (Lee & Park, 2024), from 1 (strongly disagree) to 5 (strongly agree). Your scores feed the decision tree and the prediction.
Your profile
Walk the decision tree
6 key splitsThe path highlights the branches your standardized scores follow. Each split uses the study's reported threshold and observation count (Table 2). Only the most informative split at each depth is shown.
Predicted AI self-efficacy
What drives AI self-efficacy
Relative importanceFeature-importance analysis ranks how much each literacy dimension reduces prediction error. Click a bar to learn what it measures.
Dimension detail
Select a dimension
Click any bar on the left to see its definition, reliability, and role in the tree.
Observed vs. predicted
R² = .532A Decision Tree Regression explains ~53% of the variance in AI self-efficacy. Your position is plotted against the model's fit line.
Personalized development
Enter your literacy profile to receive targeted recommendations.