Metaanálisis de generalización de la fiabilidad de los mitos de violación / Meta-analysis of reliability generalization of the rape myths

RESUMEN

Resumen Los mitos de violación son actitudes y creencias generalmente falsas, amplias y persistentes, acerca de la violación, la víctima y el agresor, que son utilizadas para negar o justificar la agresión sexual hacia las mujeres. En las últimas dos décadas, los instrumentos más utilizados para medir este constructo corresponden a la escala de aceptación de mitos de violación de Illinois (IRMAS), que utiliza expresiones directas y explícitas mediante un lenguaje clásico, y la escala de aceptación de mitos modernos de agresión sexual (AMMSA) que usa un lenguaje sutil, indirecto y moderno. Se realizó un metaanálisis de generalización de la fiabilidad de 69 estudios empíricos que utilizaron alguna de las dos escalas de mitos de violación. El objetivo fue estimar la fiabilidad media de las puntuaciones combinadas de las escalas IRMAS y AMMSA para obtener un valor aproximado de su fiabilidad general y evaluar el posible efecto moderador de algunas variables de interés. El promedio de la fiabilidad por consistencia interna de las puntuaciones de las escalas para las 98 muestras estudiadas fue de .85, IC95 % [.84, .86]. Se observó una alta heterogeneidad (I. = 96 %), y el número de ítems es la única variable moderadora que explica significativamente la variabilidad de la fiabilidad observada. Estos resultados muestran que ambas escalas presentan índices de consistencia interna aceptables en sus diversas aplicaciones. Por lo tanto, las medidas de aceptación de mitos de violación cumplen con los criterios de fiabilidad adecuados para ser utilizadas en investigaciones empíricas en distintos contextos.

ABSTRACT

Abstract Rape myths are widespread and persistent attitudes, beliefs, and stereotypes, usually false, about rape, the victim, and the perpetrator. Their function is to deny and justify sexual assaults against women, affecting the victim's attributions of responsibility and the perpetrator's attributions of guilt in rape cases. These myths exert a bias in the processing of information, directing attention and perception toward stimuli that justify the victim's responsibility for sexual aggression. These beliefs can be grouped into several types of myths Myths that hold the victim responsible by arguing that women should be careful and not expose themselves to avoid sexual aggression, myths that justify and reduce the responsibility of the aggressor by stating that the man could not contain his sexual desire and those myths that deny or normalize sexual aggression, which propose that rape occurs only in very specific contexts. In the last two decades, the instruments most commonly used to measure these beliefs are The Illinois Rape Myth Acceptance Scale (IRMAS), which uses direct and explicit expressions through classic language, and the Modern Sexual Assault Myth Acceptance Scale (AMMSA), where its expressions are modern, subtle and indirect. Considering the wide use of these instruments, it is justified to provide empirical evidence showing information on the psychometric properties of these scales. One of the procedures for synthesizing empirical results is meta-analyses (MA). This methodology can synthesize studies of specific variables and analyze the psychometric properties of the measurement instruments, providing relevant information on the quality of a given scale. Within this last type of RM are reliability generalizations (RG), those that study the reliability coefficients obtained in different applications of a scale, providing evidence on the properties of the measures used in measuring a construct. A meta-analysis of the RGs of 69 empirical studies that used any of the rape myth scales was performed. The objective was to estimate the mean reliability of the combined scores of the IRMAS and AMMSA scales to obtain an approximate value of their overall reliability and to assess the possible moderating effect of some variables of interest (e.g., research design, culture, sample type, etc.). The mean internal consistency reliability of the scale scores for the 98 samples studied was .85, 95 % C.I. [.84, .86] and the mean coefficient for each of the IRMAS and AMMSA scales was .84 and .85 respectively. All these values are above .80, a value established as satisfactory reliability of the instrument for general research. The Cronbach's alpha coefficients reported by the studies ranged from .71 to .98, with values considered moderate to excellent. These results show that both scales present acceptable internal consistency indices in various applications. There is high heterogeneity (I. = 96 %), with the number of items being the only moderating variable significantly explaining the observed reliability variability. This result was to be expected, given that the effect of test length on the estimation of reliability indices has a long tradition and is widely known in the psychometric literature.