UNIDAD DE INVESTIGACIÓN:
DISEÑO Y ANÁLISIS EN PSICOLOGÍA APLICADA
(Designs in Applied Psychology)
Universitat de València
ÚLTIMA ACTUALIZACIÓN 27 DE ENERO DE 2015
TAMAÑO DEL EFECTO
Ø ¿Qué es un tamaño del efecto? Robert Coe. Centre for Evaluation & Monitoring (CEM).
Ø Concepto. Lee A. Becker. University of Colorado.
PROGRAMAS DE CÁLCULO
Ø Descarga un programa que ejecuta el cálculo del tamaño del efecto en Excel 5/95 y sus intervalos de confianza. Robert Coe. Centre for Evaluation & Monitoring (CEM).
Ø Effect Size Calculators. Calcula on line el tamaño del efecto d de Cohen y el de correlación rYl utilizando medias y desviaciones típicas y valores t para grupos independientes con sus grados de libertad. Lee A. Becker. University of Colorado.
Ø Effect Size Calculators. Calculate a standardized mean difference (d) (Cohen, Glas and Hedges) and the strength of association (r and r2).
Ø Computation of Effect Sizes. Comparison of groups with equal size (Cohen's d and Glass Δ), Comparison of groups with different sample size (Cohen's d, Hedges' g) Effect size for mean differences of groups with unequal sample size within a pre-post design (dcorr sensu Klauer, 2001), Calculation of d and r from the test statistics of dependent and independent t-tests, Computation of d from the F-value of Analyses of Variance (ANOVA), Calculation of effect sizes from ANOVAs with multiple groups, based on group means, Increase of intervention success: The Binomial Effect Size Display (BESD) and Number Needed to Treat (NNT), Risk Ratio, Odds Ratio and Risk Difference, Effect size for the difference between two correlations, Computation of the pooled standard deviation, Transformation of the effect sizes d, r, f, Odds Ratio and η2, Computation of the effect sizes d, r and η2 from χ2- and z test statistics, Table of interpretation for different effect sizes.
EFFECT SIZE CALCULATION (estimación)
Whereas statistical significance tests assess the reliability of the relationship between independent and dependent variables, effect sizes assess the strength of the relationship. In general, you need to know the effect size you hope to achieve to calculate statistical power.
Effect size can be measured as the standardized difference between two means, or as the correlation between the independent variable classification and the individual scores on the dependent variable, referred to as the effect size correlation.
Cohen defined d as the difference between the means, M1-M2, divided by the standard deviation of either group. For example, the groups in your study could refer to the experimental and control groups. The standard deviation of either group in your study can be used when the variances of the two groups are homogeneous.
Effect sizes are generally defined as small (d = 0.2), medium (d = 0.5), and large (d = 0.8).
Several formulas could be used to calculate effect size. According to Cohen:
d = M1-M2 / Ö [( s1² + s 2²) / 2]
d = M1-M2 / s, where s = Ö [å (X - M)² / N]
In this case X is the raw score, M is the mean, and N is the number of cases.
F-test ANOVA Effect Size
For ANOVA, the effect size index f is used. You can compute the effect size index from the group means. Effect sizes are generally defined as small (f = 0.1), medium (f = 0.25), and large (f = 0.4).
Correlations Effect Size
To evaluate the null hypothesis that a product moment correlation in the population is zero (r = 0). The effect size symbol is r. Effect sizes are generally defined as small (r = 0.1), medium (r = 0.3), and large (r = 0.5).
Regression Effect Size
Effect size for regression reflects the variance accounted for by some source in the population relative to the residual variance proportion (η2). Effect sizes are generally defined as small (η2 = 0.02), medium (η2 = 0.15), and large (η2 = 0.35).
Documentación sobre tamaño del efecto y análisis estadístico
· Online Statistics Education: An Interactive Multimedia Course of Study
TAMAÑO DEL EFECTO: EJEMPLOS DE APLICACIONES
· Presentación de un ejemplo: Kirsch, I., & Sapirstein, G. (1998). Listening to Prozac but Hearing Placebo: A Meta-Analysis of Antidepressant Medication. Prevention & Treatment, Volume 1, Article 0002a, posted June 26, 1998.