Project Title and Abstract
Identifying When Mindset Interventions are Effective for Students: A Two-Model Machine Learning Approach
Mindset has an important influence on learning; furthermore, research has shown that interventions to change mindset can have a positive effect. However, little is known about the student- and school-level characteristics (e.g., math anxiety, sense of belonging) that influence whether or not a mindset intervention will be beneficial for learning. We propose an exploratory approach to identify which student and school characteristics are crucial for predicting intervention efficacy, by training two machine learning models on data from the National Study of Learning Mindsets. We will train one model within the control condition to predict longitudinal grade improvement from student characteristics, thereby also revealing which characteristics (including multivariate combinations) are most predictive. The second model will be trained on experimental condition data to predict the difference between expected grade improvement (found by applying the first model) and actual grade improvement, thereby revealing which characteristics predict intervention efficacy.