The Mediational Role of Math Anxiety in the Link Between Loneliness and Math Achievement: An Analysis using PISA 2022

There is mounting evidence on how non-cognitive skills (or psychosocial resources) play a crucial role in supporting students throughouttheir educational journey and are crucial for social, scholastic, and personal adjustment (Heckman & Rubinstein, 2001). However,compared to other countries, this topic has received little attention from Italian researchers (Pisanu & Fraccaroli, 2022). Similarly, Italianpolicy also lags in paying attention to such evidence. Only recently have some ministerial reforms sought to promote the knowledge,application, and discussion of supporting non-cognitive skills in Italian educational settings. In this study, we delve into non-cognitive skills,focusing on the underestimated role of loneliness (i.e., the unpleasant feeling that emerges when individuals perceive their social networkas qualitatively and/or quantitatively scarce; Asher & Paquette, 2003). It has been shown that loneliness is negatively associated withmental health and positively correlated with symptoms of anxiety and depression (Houghton et al., 2022). Furthermore, a recent study bySette et al. (2023) found a negative link between loneliness and students’ positive academic self-perceptions. However, the literature stilloffers few studies on the associations between loneliness, psychosocial resources, and objective measures of academic achievement. Therefore, in this project, we examined the link between loneliness, math anxiety, and academic achievement in mathematics among 15-year-old Italian adolescents using the PISA 2022 survey data. Drawing on previous literature on the effect of loneliness on academic self-perceptions, we hypothesized and tested a model in which math anxiety mediates the relationship between loneliness and mathachievement. We analyzed data from 10552 Italian students (50.8% girls). Structural Equation Models were used to test our hypothesizedmodel. In addition, covariates (i.e., gender, indices of economic, social, and cultural status, cooperation, experiences of being bullied,quality of student-teacher relationships, and expected occupation status) were included in the analysis. To operationalize loneliness, weused a latent variable composed of three items from the sense of belonging scale consistent with the items in a widely used lonelinessquestionnaire (i.e., Asher & Wheeler, 1985). Math anxiety was measured by a latent variable composed of 6 items. Finally, mathachievement was measured through a single indicator, which reflected the “Plausible Value 1” in PISA 2022. The model fitted the data wellaccording to commonly used criteria (CFI and TLI > .90; RMSEA < .06; SRMR < .08). Concerning parameters of interests, we found that(a) loneliness significantly and positively affected math anxiety, and the latter significantly and negatively affected math achievement; (b)the direct effect of loneliness on math achievement was not significant (albeit the latent zero-order correlation was negative and significant)and constraining this parameter to be zero did not significantly worse the fit of the model; (c) bootstrap analysis (5000 resamplings)showed that the 95% bias-corrected confidence interval of the indirect effect was significantly different from 0. This analysis provides a firstattempt to empirically unravel the relationship between two underestimated non-cognitive aspects (loneliness and math anxiety) and theirimpact on math achievement, thus offering potential insights for future longitudinal studies and practical interventions at school.