Predicting Success from Day One

This research examines non-academic predictors of student success in the Faculty of Engineering at Stellenbosch University. Using data from 532 students in the 2021 cohort tracked over four years, we applied rigorous statistical methods including False Discovery Rate correction to identify reliable wellbeing factors that distinguish between degree completers, unsuccessful persisters, and leavers. Our analysis of 48 wellbeing measures across 80+ demographic scenarios reveals that physical health, life satisfaction, financial security, willpower, hope, and social support serve as critical intervention targets, with effects varying significantly across gender, socioeconomic background, generational status, and academic preparation levels. The findings establish universal foundations for engineering student success while demonstrating that different demographic groups require targeted interventions based on distinct patterns of wellbeing predictors. This work provides evidence-based guidance for implementing prevention-focused support systems that could improve student success rates while optimizing resource allocation across diverse engineering student populations.

Can wellbeing measures taken at university entry really predict academic outcomes years later?

This presentation addresses a fundamental question underlying all student success research: the temporal stability of non-academic measures. Using extensive longitudinal research evidence, we examine whether the 49 wellbeing, mindset, and behavioral factors in our SUBSIFY framework demonstrate sufficient stability over time to enable meaningful long-term prediction. Key findings reveal that psychological constructs show exceptional stability, wellbeing measures demonstrate strong temporal consistency, and even behavioral measures maintain meaningful stability over multi-year periods. Importantly, these patterns persist naturally but remain malleable through targeted intervention. Without temporal stability, early identification would be impossible—with it, we can confidently predict student trajectories and intervene early with evidence-based support to change outcomes before problems emerge.

The results presented on this page are based on data collected through SUBSIFY (Stellenbosch University Baseline Survey for Incoming First-Years), a comprehensive assessment framework that measures 49 validated factors across four major categories: Flourishing Index Components (20 factors), Wellness Habits (14 factors), EPOCH Well-being Measures (6 factors), and Mindset Measures (9 factors). To better understand what these wellbeing factors represent and how they are measured, watch the introductory video below, which provides a detailed overview of each category and the specific measures within SUBSIFY. This framework goes beyond traditional academic predictors to capture the broader range of non-academic factors that may influence student success, providing the foundation for the faculty-specific analyses presented here.

This video presents a comprehensive examination of significant achievement gaps within the Faculty of Engineering, revealing stark disparities that demand immediate attention and action. Through rigorous analysis of 532 students from our SUBSIFY survey tracked longitudinally, we document a 29.8 percentage point gap between the highest and lowest performing demographic groups, with DCM success rates ranging from 47.8% for White students to 18.0% for Black African students. The presentation systematically examines how race, academic preparation, socioeconomic status, and generational status intersect to create compounded barriers for vulnerable student populations. Mathematics preparation emerges as the most critical predictor, showing a dramatic 56.8 percentage point gap between highest and lowest performers. Moving beyond simply documenting these disparities, the video demonstrates how our wellbeing research offers evidence-based pathways to closing these gaps through targeted interventions that address modifiable factors like physical health, willpower, and social support. This analysis represents one of the first systematic examinations of wellbeing predictors across demographic groups in South African engineering education, providing both sobering documentation of existing inequalities and actionable solutions for creating more equitable pathways to student success.

This video demonstrates the exact statistical procedures we followed throughout our entire study, making it essential viewing for readers who want to understand how we arrived at our main findings or need clarification on our methodological approach. Through an analysis of 4,224 individual t-tests across 88 demographic scenarios involving 532 engineering students tracked from enrollment to graduation, we reveal how advanced statistical techniques—including False Discovery Rate correction and effect size analysis—distinguish genuine predictive factors from statistical noise. Using willpower as our primary example, the presentation walks viewers through the complete analytical process from raw survey data to evidence-based intervention recommendations, showcasing how small but consistent effects (Cohen's d = 0.43-0.62) translate into meaningful population-level improvements in engineering student outcomes. This rigorous methodology established willpower as a Priority #1 universal intervention target and serves as a template for identifying reliable wellbeing predictors that can guide institutional decision-making with statistical confidence.

To ensure transparency and scientific rigor in our student success research, we have developed a comprehensive statistical framework that examines wellbeing factors across diverse demographic groups and student pathways. The analysis you see reported for this faculty employs sophisticated multiple comparison corrections, effect size calculations, and variance testing to distinguish genuine predictive factors from statistical noise. This methodology video provides detailed insight into the four-step analytical process we use, explaining techniques such as Levene's test for variance assumptions, False Discovery Rate correction using the Benjamini-Hochberg procedure, and Cohen's d effect size interpretation. Understanding these statistical foundations is crucial for interpreting our findings correctly - particularly why we report both uncorrected and FDR-corrected results, and how our three-pathway student model (degree completers, unsuccessful persisters, and leavers) provides more nuanced insights than traditional binary success/failure approaches. The rigorous application of these methods across thousands of statistical tests ensures that the intervention targets we identify represent reliable, evidence-based opportunities for supporting student success rather than chance findings.

Note: This video was originally created for our EMS faculty analysis, but all methodological questions and implementation guidance apply equally to Engineering students. Please disregard any specific EMS references, as the underlying statistical approaches, intervention principles, and implementation strategies are identical across both faculties.

This video addresses the most frequently asked questions about our comprehensive wellbeing analysis of student success patterns. The presentation covers critical methodological questions including why we used False Discovery Rate correction to ensure reliable results, how our 4,224 statistical comparisons were generated across demographic scenarios, and what effect sizes mean for practical intervention planning. It also addresses implementation questions such as which interventions to prioritize first, cost-effective funding strategies, and how findings might transfer to other faculties. Most importantly, the video provides an honest assessment of study limitations, particularly the distinction between correlation and causation, and outlines the intervention research needed to prove that improving wellbeing factors actually causes better student outcomes. Whether you're interested in the statistical rigor behind our findings or practical guidance for implementing evidence-based student support programs, this Q&A provides essential context for understanding how these research results can inform institutional decision-making.

We are currently developing academic paper(s) that will unpack this research, its results, and implications in detail. These publications will provide comprehensive scholarly analysis of the wellbeing factors that predict student success in the Faculty of Engineering. The papers will be made available on this page once they become accessible.