This project introduces the concept of effect size as a critical statistical measure used to determine the magnitude of relationships or differences between variables. It focuses on four commonly used effect size metrics—Cohen's d, eta squared (η²), partial eta squared (ηp²), and Pearson’s correlation coefficient (r). Through practical examples and data analysis, the project highlights how these measures provide deeper insights into the practical significance of research findings, complementing traditional hypothesis testing.
-
To calculate and interpret various effect size measures—including Cohen’s d, eta squared, partial eta squared, and Pearson’s r—using real or simulated datasets.
-
To evaluate the practical significance of statistical findings by comparing and applying appropriate effect size metrics in different analysis contexts.
The project begins by selecting appropriate datasets suitable for comparing group differences and examining relationships between variables. Descriptive statistics are first calculated to understand the data distribution. For mean comparisons, effect sizes such as Cohen’s d, eta squared, and partial eta squared are computed using results from t-tests and ANOVA. Pearson’s correlation coefficient is used to measure the strength and direction of linear relationships between continuous variables. All analyses are conducted using Python and relevant statistical libraries such as scipy, pingouin, and statsmodels, with results interpreted based on established thresholds for small, medium, and large effects.