Prediction Versus Explanation in Educational Psychology: a Cross-Theoretical Approach to Using Teacher Behaviour to Predict Student Engagement in Physical Education

Abstract

Educational psychology usually focuses on explaining phenomena. As a result, researchers seldom explore how well their models predict the outcomes they care about using best-practice approaches to predictive statistics. In this paper, we focus less on explanation and more on prediction, showing how both are important for advancing the field. We apply predictive models to the role of teachers on student engagement, i.e. the thoughts, attitudes, and behaviours, that translate motivation into progress. We integrate the suggestions from four prominent motivational theories (self-determination theory, achievement goal theory, growth mindset theory, and transformational leadership theory), and aim to identify those most critical behaviours for predicting changes in students’ engagement in physical education. Students (N = 1324 all from year 7, 52% girls) from 17 low socio-economic status schools rated their teacher’s demonstration of 71 behaviours in the middle of the school year. We also assessed students’ engagement at the beginning and end of the year. We trained elastic-net regression models on 70% of the data and then assessed their predictive validity on the held-out data (30%). The models showed that teacher behaviours predicted 4.39% of the variance in students’ change in engagement. Some behaviours that were most consistently associated with a positive change in engagement were being good role models (β = 0.046), taking interest in students’ lives outside of class (β = 0.033), and allowing students to make choices (β = 0.029). The influential behaviours did not neatly fit within any single motivational theory. These findings support arguments for integrating different theoretical approaches, and suggest practitioners may want to consider multiple theories when designing interventions. More generally, we argue that researchers in educational psychology should more frequently test how well their models not just explain, but predict the outcomes they care about.

The Effectiveness of Virtual Tour Museums on Student Engagement in Social Studies Learning in Elementary Schools

“…Additionally, the student engagement level in the experimental class was classified as high. The utilization of real-life illustrations in social studies education has been shown to be a successful approach in fostering comprehension and achieving concept mastery (Kassab et al, 2023;Noetel et al, 2023;Yang et al, 2023). The use of this approach enhances student engagement and facilitates comprehension of the offered content.…”

How the Predictors of Math Achievement Change over Time: A Longitudinal Machine Learning Approach

“…PREDICTING MATH ACHIEVEMENT OVER TIME 9 Recently, there have been a growing number of emerging studies that have utilized machine learning methods to predict achievement outcomes from cognitive and non-cognitive skills, with the goal to better understand important psychological predictors (FİLİZ & Ersoy, 2020;Gamazo & Martínez-Abad, 2020;Kiray et al, 2015;Martinez Abad & Chaparro Caso López, 2017;Nadaf et al, 2021;Nadaf et al, 2023;Noetel et al, 2023;Yoo, 2018). For example, Yoo (2018) conducted an Elastic Net regression analysis to predict Korean 4th graders' math achievement scores with more than 150 predictor variables.…”

Breaking down silos in motivation science.

“…When other constructs are considered, they are often treated as competitors in empirical work designed to test which construct accounts for more variance. This focus on uniqueness and discriminant validity certainly has its place, in that it helps combat construct proliferation and redundancy (Hattie et al, 2020; Noetel et al, 2023). However, it can easily be overemphasized, with the result being a divisiveness and an “us versus them” mentality that bolsters and reinforces the silos (Anderman, 2020; Elliot & Sommet, 2023; Pekrun, 2024).…”