Calculating the main alternatives to null-hypothesis-significance testing in between-subject experimental designs / Cálculo de las principales alternativas a la prueba de significación de la hipótesis nula en diseños experimentales intersujetos

The present article deals with the controversy about null hypothesis significance testing (NHST) that currently exists in psychological research. The main criticisms and counter-criticisms posed by its detractors and supporters are presented. Alternatives have been proposed to replace or complement the NHST, as recommended by the Task Force on Statistical Inference of the APA, and these are explained. Furthermore, the procedure for calculating these alternatives using SPSS statistical software in a twoway between-factor design is demonstrated. Thus, an attempt is made to provide the applied researcher with resources that make it possible to analyse and interpret the results of any research study using a group of indicators that lends a high level of validity to the statistical inference performed (AU)

El presente artículo se centra en la controversia existente actualmente en torno a la prueba de significación de la hipótesis nula en la investigación psicológica. Se exponen las principales críticas y contra-críticas que plantean sus detractores y defensores, así como las alternativas que, contando con el apoyo de la Task Force on Statistical Inference de la APA, han sido propuestas para sustituirla o complementarla. Además, se muestra la forma en la que pueden calcularse, mediante el programa estadístico SPSS, tales índices alternativos en un diseño factorial aleatorio. Con ello se pretende dotar al investigador aplicado de una serie de recursos que le permitan analizar e interpretar los resultados de cualquier estudio manejando un conjunto de indicadores que proporcionan un alto grado de validez a la inferencia realizada (AU)

Calculating the main alternatives to null-hypothesis-significance testing in between-subject experimental designs.

The present article deals with the controversy about null hypothesis significance testing (NHST) that currently exists in psychological research. The main criticisms and counter-criticisms posed by its detractors and supporters are presented. Alternatives have been proposed to replace or complement the NHST, as recommended by the Task Force on Statistical Inference of the APA, and these are explained. Furthermore, the procedure for calculating these alternatives using SPSS statistical software in a two-way between-factor design is demonstrated. Thus, an attempt is made to provide the applied researcher with resources that make it possible to analyse and interpret the results of any research study using a group of indicators that lends a high level of validity to the statistical inference performed.

Manipulación del contenido en el fenómeno de ensombrecimiento en juicios de contingencia / Manipulation of the content in the overshadowing in contingency judments

En el presente artículo pretendemos estudiar el proceso de ensombrecimiento y la predicción que realizan los modelos establecidos en los juicios de contingencia para explicar este fenómeno. En dos experimentos, los sujetos establecieron sus juicios en una tarea experimental que consistía en predecir dos consecuencias en función de cuatro causas potenciales. Los dos experimentos diferían en cuanto al contenido de las causas y consecuencias (síntomas y enfermedades en el primero y letras y números en el segundo). Los resultados muestran que el contenido parece no afectar a los juicios aunque sí hay diferencias entre ellos en función de las contingencias objetivas establecidas entre las causas y consecuencias. De los modelos estudiados sólo tres de ellos muestran ajustes aceptables a los datos empíricos: Regla de Bayes, Delta o y el Modelo Asociativo (AU)

This article studies the process of overshadowing and the predictions realized by the models of contingency judgments. In two different experiments, subjects were provided with four cues, to which they had to respond in one of two possible outcomes. The two experiments differed with respect to the content of their cues and outcomes: symptoms and diseases in the first experiment, letters and numbers in the second one. The results show that content does not seem to affect the subjects’ judgment. However, the results do differ with respect to the contingencies established between the cues and the subjects’ outcomes. Of all the models studied, only three proved to acceptably fit the empirical data: the Bayes’ rule, the Delta O, and the Associative Model (AU)