Predicting academic success: The roles of emotions, stress and prior grades

Emotional regulation, anxiety, perceived academic stress and prior grades each predict academic success, but existing research treats them as independent predictors rather than testing whether their effects converge through a shared emotional pathway, a gap most consequential in high-pressure settings where cognitive and affective demands rise together, and additive models are least likely to capture how achievement is actually produced. Control-value theory and social cognitive theory jointly motivate a more integrated test: control-value theory holds that appraisals of control generate achievement emotions, including anxiety, that shape performance, while social cognitive theory treats prior grades as a mastery experience calibrating self-efficacy and emotional vulnerability. Together, the frameworks predict that anxiety and perceived academic stress should mediate emotional regulation and prior grades through to academic achievement, yet this integrated model has rarely been empirically tested and never in Indian engineering education, a high-pressure, non-Western setting where these demands converge most acutely. The present study tested academic anxiety and perceived academic stress as mediators of the effects of emotional regulation and prior grades on academic achievement among undergraduate engineering students in India, using a PLS-SEM approach.

Data from 500 undergraduate engineering students enrolled at private institutions in India were collected via a structured online survey. Validated scales measured emotional regulation (ERQ), anxiety (GAD-7), and perceived academic stress (PAS). Prior grades and current academic achievement were measured using self-reported grade point averages (GPAs). Hypothesised relationships were tested using Partial Least Squares Structural Equation Modelling (PLS-SEM) with SmartPLS 4, including bootstrapped mediation analysis with 10,000 resamples.

Prior grades emerged as the strongest direct predictor of academic achievement (β = 0.388, p < .001) and were negatively associated with anxiety (β = -0.155, p < .001). Emotional regulation positively predicted anxiety (β = 0.168, p < .001) and perceived academic stress (β = 0.134, p = .001) while negatively predicting academic achievement (β = -0.121, p = .001). Anxiety negatively predicted academic achievement (β = -0.084, p = .048) and was the strongest predictor of perceived academic stress (β = 0.554, p < .001). Perceived academic stress did not directly predict academic achievement. Anxiety significantly mediated the effect of emotional regulation on academic achievement (β = -0.014, 95% CI [-0.033, -0.001]) and the effect of prior grades on academic achievement (β = 0.013, 95% CI [0.002, 0.030]); pathways through perceived academic stress alone and sequential mediation chains did not reach significance.

Anxiety, not perceived academic stress, emerged as the central mechanism linking both emotional regulation and prior grades to academic achievement, consistent with control-value theory's account of anxiety as an appraisal-driven mediator and social cognitive theory's account of prior mastery experiences as a source of emotional vulnerability. This boundary finding refines both theories: perceived academic stress functioned as a downstream outcome of anxiety rather than an independent or sequential mediator, and emotional regulation was not uniformly protective, predicting heightened anxiety and lower achievement rather than buffering either, consistent with maladaptive strategies predominating under sustained pressure. Practically, the results suggest interventions built on regulation capacity alone, or on managing perceived academic stress directly, are likely to miss the upstream anxiety pathway that actually links prior grades and regulation to achievement in this population. These findings provide the first empirical test of this integrated model in Indian engineering education, and the first evidence that its sequential pathway does not hold in this context.