24.Estad_ Tamaños del efecto _mayo2025

Understanding Effect Sizes in Statistical Analysis

Conceptualization and Importance of Effect Sizes

• The discussion begins with the need to conceptualize and value effect sizes, especially after addressing controversies surrounding statistical significance in previous lessons.

• Key questions arise regarding the existence of a phenomenon, its magnitude, and the importance of that magnitude when interpreting results.

• Inferential analysis is introduced as a method to determine whether a phenomenon exists, highlighting debates on the utility of statistical significance.

• The speaker emphasizes questioning the relevance of discussing null hypotheses concerning variable relationships and how this leads to focusing on effect sizes for interpretation.

• A modern definition of effect size is presented as a reflection of what occurs within data sets, contrasting it with other terms like "magnitude."

Levels of Understanding Effect Size

• The speaker outlines three levels related to understanding effect size:

• A formula for measurement,

• The logic behind comparisons (e.g., group means),

• The resulting numerical magnitude from these calculations.

• Preference for using "effect size" over "magnitude" is justified by translation accuracy and established terminology in research literature.

Relevance and Reporting Standards

• An important editorial from the Journal of Applied Psychology stresses that reporting effect sizes is essential in research publications.

• Discussion includes the formation of a task force by APA's scientific affairs committee to address controversies around statistical significance testing in psychology journals.

• Notable figures involved include experts from various fields such as statistics, psychology, and education who contributed insights into best practices for reporting findings.

Historical Context and Guidelines

• The historical context reveals that guidelines emphasizing effect size reporting were established as early as 1999, with further reinforcement in subsequent APA manuals published later.

• Critiques are noted regarding defects in report designs if they fail to include effect sizes, indicating an evolving standard towards more rigorous statistical reporting practices.

Effect Sizes in Statistical Analysis

Understanding Effect Sizes

• The concept of effect sizes is emphasized, noting that they are crucial to report alongside significance tests. The null hypothesis's relevance depends on the editor, but effect sizes remain important.

• Various formulas for comparing groups are introduced, including the common language effect size and probability of superiority. However, the focus will be on standardized mean differences as a common method.

• Standardized mean differences involve comparing means divided by standard deviation. Correlation coefficients also serve as an effect size measure in correlation analysis.

Valuing Effect Sizes

• The valuation of effect sizes varies by discipline and practice area. Cohen provides guidelines for interpreting these values based on context.

• Suggested cut-off points for correlation strength: small (0.10), moderate (0.30), and large (0.50). Values below these thresholds may be considered irrelevant.

Personal Anecdote and Evolution of Understanding

• A personal anecdote from 2013 highlights challenges faced when researching effect sizes during thesis work, particularly due to limited resources available in Spanish.

• The evolution of teaching methodologies is noted; today, undergraduate courses commonly include instruction on effect sizes, making them more accessible than before.

Conclusion and Next Steps

• Acknowledgment that there are multiple ways to assess effect sizes beyond Cohen’s suggestions encourages critical evaluation of different approaches.

• Invitation to the next lesson focuses on presenting results effectively while incorporating discussions about reporting effect sizes accurately.