A while ago I promised a longer post on standardized effect size. This isn't it. Instead it is a link to my piece in the February 2010 issue of The Psychologist.
I had intended to write a short summary of my 2009 BJP paper, but that didn't really get off the ground. However, I had for a while wanted to write about the Society for the Suppression of the Correlation Coefficient. I had first read about this in Tukey's (1954) chapter on regression and path analysis. This is one of the earlier papers in the literature criticizing the preference for (standardized) correlation coefficients over (simple, unstandardized) regression coefficients. Read the article for an earlier example! I was reminded of its existence by Brillinger's (2001) paper.
For non-members of the BPS the link below contains a pre-publication version of 'When correlations go bad ...'.
I may also get around to the web site one day.
P.S. I adapted data and R code from Gelman and Hill (2007) for the example. I chose it because it a nice simple example of regression and because it is also a pointer to someone who argues in favour of standardization )at least in some situations). Both my 2009 paper and the 'When correlations go bad ...' are my attempts at getting people to rethink the use of correlation and standardization. The tone is deliberately (slightly?) polemic. There are other views so don't just take my word for it ... think about what you are trying to do and decide whether a correlation coefficient (or a standardized mean difference) is the right option.
P.S. I adapted data and R code from Gelman and Hill (2007) for the example. I chose it because it a nice simple example of regression and because it is also a pointer to someone who argues in favour of standardization )at least in some situations). Both my 2009 paper and the 'When correlations go bad ...' are my attempts at getting people to rethink the use of correlation and standardization. The tone is deliberately (slightly?) polemic. There are other views so don't just take my word for it ... think about what you are trying to do and decide whether a correlation coefficient (or a standardized mean difference) is the right option.
References
Baguley, T. (2009). Standardized or simple effect size: What should be reported? British Journal of Psychology. 100, 603–617.
Baguley, T. (2010). When correlations go bad … The Psychologist, 23, 122-3.
Brillinger, D.R. (2001). John Tukey and the correlation coefficient, Computing Science and Statistics, 33, 204–218.
Gelman, A. & Hill, J. (2007). Data Analysis Using Regression and Multilevel/Hierarchical Models (Analytical Methods for Social Research)
. Cambridge: Cambridge University Press.
Tukey, J.W. (1954). Causation, regression and path analysis. In O. Kempthorne, T.A. Bancroft, J.W. Gowen & J.L. Lush (Eds.) Statistics and mathematics in biology (pp.35–66). Ames, IA: Iowa State College Press.
Baguley, T. (2010). When correlations go bad … The Psychologist, 23, 122-3.
Brillinger, D.R. (2001). John Tukey and the correlation coefficient, Computing Science and Statistics, 33, 204–218.
Gelman, A. & Hill, J. (2007). Data Analysis Using Regression and Multilevel/Hierarchical Models (Analytical Methods for Social Research)
. Cambridge: Cambridge University Press.Tukey, J.W. (1954). Causation, regression and path analysis. In O. Kempthorne, T.A. Bancroft, J.W. Gowen & J.L. Lush (Eds.) Statistics and mathematics in biology (pp.35–66). Ames, IA: Iowa State College Press.
No comments:
Post a Comment