See this thread on the SPM mailing list:

[It should be said] that in SPM (and generally) we discourage the use of correlations – they are difficult to interpret in many instances and are an odd measure of (standardised) effect size that conflates the regression slope with standard error. In fact, there are people who would like to abolish the correlation coefficient [...]: http://societytosupressthecorrelationcoefficient.wordpress.com/about/ I am not sure whether this society is serious but it raises an interesting point of view :) |