Trends in statistical analysis: The frequentist inference limitations and the possibilities of the Bayesian approach.
Statistical procedures used in psychological research are in full review. The accumulation of criticism upon classical techniques has led to different answers. On one hand, there is a deepening on pushing analysis beyond hypothesis testing and more elaborate interpretation of data. On the other hand...
Main Author: | |
---|---|
Format: | Online |
Language: | spa |
Published: |
Instituto de Investigaciones Psicológicas (IIPSI, Conicet-UNC)
2011
|
Subjects: | |
Online Access: | https://revistas.unc.edu.ar/index.php/revaluar/article/view/2842 |
_version_ | 1811171852608864256 |
---|---|
author | Bologna, Eduardo Bologna, Eduardo |
author_facet | Bologna, Eduardo Bologna, Eduardo |
author_sort | Bologna, Eduardo |
collection | Portal de Revistas |
description | Statistical procedures used in psychological research are in full review. The accumulation of criticism upon classical techniques has led to different answers. On one hand, there is a deepening on pushing analysis beyond hypothesis testing and more elaborate interpretation of data. On the other hand, more and more researchers suggest introducing Bayesian approach to make inferences, and there are more software routines to facilitate this. On the frequentist approach, this paper discusses additional procedures to test hypotheses suggested in the sixth edition APA Publication Manual. We review the rationale of hypothesis testing; we show the usefulness of reporting effect size measures ―and the difficulty to infer about them―, power analysis, and the different alternatives to calculate confidence intervals. On the Bayesian approach, we discuss the assumptions and the advantages in terms of information it offers and its contribution to the cumulative nature of the results. Then we show a simple illustrative application of its rationale, comparing the estimation of proportions by means of credibility and confidence intervals. |
format | Online |
id | oai:ojs.revistas.unc.edu.ar:article-2842 |
institution | Universidad Nacional de Cordoba |
language | spa |
publishDate | 2011 |
publisher | Instituto de Investigaciones Psicológicas (IIPSI, Conicet-UNC) |
record_format | ojs |
spelling | oai:ojs.revistas.unc.edu.ar:article-28422021-07-05T22:57:34Z Trends in statistical analysis: The frequentist inference limitations and the possibilities of the Bayesian approach. Tendencias en el análisis estadístico: Límites de la inferencia frecuencial y posibilidades del enfoque bayesiano Bologna, Eduardo Bologna, Eduardo inferencia frecuencial inferencia bayesiana prueba de hipotesis estadistica en psicologia Statistical procedures used in psychological research are in full review. The accumulation of criticism upon classical techniques has led to different answers. On one hand, there is a deepening on pushing analysis beyond hypothesis testing and more elaborate interpretation of data. On the other hand, more and more researchers suggest introducing Bayesian approach to make inferences, and there are more software routines to facilitate this. On the frequentist approach, this paper discusses additional procedures to test hypotheses suggested in the sixth edition APA Publication Manual. We review the rationale of hypothesis testing; we show the usefulness of reporting effect size measures ―and the difficulty to infer about them―, power analysis, and the different alternatives to calculate confidence intervals. On the Bayesian approach, we discuss the assumptions and the advantages in terms of information it offers and its contribution to the cumulative nature of the results. Then we show a simple illustrative application of its rationale, comparing the estimation of proportions by means of credibility and confidence intervals. Los procedimientos estadisticos que se usan en investigacion en Psicologia se hallan en plena revision. La acumulacion de criticas hacia las tecnicas clasicas ha suscitado respuestas diferentes. Por un lado, profundizar los analisis mas alla de las pruebas de hipotesis y dar mas detalles a la lectura de los datos. Por otro lado, cada vez son mas los partidarios de introducir el enfoque bayesiano al hacer inferencias y crecen las rutinas informaticas para facilitarlo. Sobre el enfoque frecuencial, en este trabajo se analizan los procedimientos adicionales a las pruebas de hipotesis, que sugiere APA en la sexta edicion del manual de publicacion. Se revisan los fundamentos logicos de las pruebas de hipotesis, se muestra la utilidad de informar medidas sobre el tamano del efecto, junto a la dificultad para inferir sobre ellas, analizar la potencia, y las diferentes alternativas para calcular intervalos de confianza. Sobre el enfoque bayesiano, se discuten los supuestos y las ventajas en cuanto a la informacion que ofrece y su aporte al caracter acumulativo de los resultados. Luego se muestra una aplicacion simple pero ilustrativa del modo en que razona este enfoque, comparando la estimacion de la proporcion a traves del intervalo de credibilidad y el de confianza. Instituto de Investigaciones Psicológicas (IIPSI, Conicet-UNC) 2011-06-01 info:eu-repo/semantics/article info:eu-repo/semantics/publishedVersion application/pdf https://revistas.unc.edu.ar/index.php/revaluar/article/view/2842 10.35670/1667-4545.v11.n1.2842 Revista Evaluar; Vol. 11 Núm. 1 (2011) 1667-4545 10.35670/1667-4545.v11.n1 spa https://revistas.unc.edu.ar/index.php/revaluar/article/view/2842/2724 Derechos de autor 2011 Eduardo Bologna http://creativecommons.org/licenses/by/4.0 |
spellingShingle | inferencia frecuencial inferencia bayesiana prueba de hipotesis estadistica en psicologia Bologna, Eduardo Bologna, Eduardo Trends in statistical analysis: The frequentist inference limitations and the possibilities of the Bayesian approach. |
title | Trends in statistical analysis: The frequentist inference limitations and the possibilities of the Bayesian approach. |
title_alt | Tendencias en el análisis estadístico: Límites de la inferencia frecuencial y posibilidades del enfoque bayesiano |
title_full | Trends in statistical analysis: The frequentist inference limitations and the possibilities of the Bayesian approach. |
title_fullStr | Trends in statistical analysis: The frequentist inference limitations and the possibilities of the Bayesian approach. |
title_full_unstemmed | Trends in statistical analysis: The frequentist inference limitations and the possibilities of the Bayesian approach. |
title_short | Trends in statistical analysis: The frequentist inference limitations and the possibilities of the Bayesian approach. |
title_sort | trends in statistical analysis the frequentist inference limitations and the possibilities of the bayesian approach |
topic | inferencia frecuencial inferencia bayesiana prueba de hipotesis estadistica en psicologia |
topic_facet | inferencia frecuencial inferencia bayesiana prueba de hipotesis estadistica en psicologia |
url | https://revistas.unc.edu.ar/index.php/revaluar/article/view/2842 |
work_keys_str_mv | AT bolognaeduardo trendsinstatisticalanalysisthefrequentistinferencelimitationsandthepossibilitiesofthebayesianapproach AT bolognaeduardo trendsinstatisticalanalysisthefrequentistinferencelimitationsandthepossibilitiesofthebayesianapproach AT bolognaeduardo tendenciasenelanalisisestadisticolimitesdelainferenciafrecuencialyposibilidadesdelenfoquebayesiano AT bolognaeduardo tendenciasenelanalisisestadisticolimitesdelainferenciafrecuencialyposibilidadesdelenfoquebayesiano |