Modeling Parenting Outcomes Using Educational Attainment, Cognitive Stimulation, and Parenting Efficacy with Machine Learning

Abstract

Objective: The present study aimed to model parenting outcomes based on educational attainment, cognitive stimulation, and parenting efficacy using machine learning techniques to determine their relative predictive power and interactions.

Methods and Materials: This study employed a cross-sectional predictive correlational design with a sample of 412 parents from Indonesia selected through stratified random sampling. Data were collected using standardized instruments, including measures of cognitive stimulation (HOME Inventory), parenting efficacy (Parenting Sense of Competence Scale), and parenting outcomes (Alabama Parenting Questionnaire), alongside a demographic measure of educational attainment. Data analysis involved both traditional statistical methods and machine learning approaches. Initial analyses were conducted using SPSS-27 to examine descriptive statistics and correlations. Subsequently, supervised machine learning models, including linear regression, random forest, support vector machine, and gradient boosting, were implemented using Python-based libraries. Model evaluation was performed using 10-fold cross-validation, and performance metrics included mean squared error (MSE), root mean squared error (RMSE), and coefficient of determination (R²). Feature importance analysis was also conducted to determine the relative contribution of predictors.

Findings: Results indicated that all predictors significantly contributed to parenting outcomes (p < 0.001), with parenting efficacy demonstrating the strongest standardized effect (β = 0.44), followed by cognitive stimulation (β = 0.38) and educational attainment (β = 0.21). The regression model explained 53% of the variance in parenting outcomes (R² = 0.53). Machine learning analyses revealed that gradient boosting achieved the highest predictive performance (R² = 0.67), outperforming random forest (R² = 0.64), support vector machine (R² = 0.59), and linear regression (R² = 0.52). Feature importance scores confirmed parenting efficacy as the most influential predictor, followed by cognitive stimulation and educational attainment.

Conclusion: The findings highlight the critical role of parenting efficacy and cognitive stimulation as primary determinants of parenting outcomes, while educational attainment exerts a significant but indirect influence. The superior performance of machine learning models underscores the complexity and nonlinear nature of these relationships, suggesting that advanced analytical approaches provide more accurate and comprehensive insights. These results have important implications for the development of targeted interventions and policies aimed at enhancing parenting practices and promoting child development.