Tim Urdan, author of Statistics in Plain English, demonstrates how to test whether two correlation coefficients are significantly different from each other.
Thanks for the vedio. It was useful. I was wondering if you can confirm whether the procedure you are using in this vedio to test the difference of coefficients of correlation for Pearson r is valid for correlation coefficients obtained from Spesrman rho as well???
Hi. No, I do not believe this procedure is valid for the Spearman rank-order correlations. The procedure in the video is for Pearson correlation coefficients, which assumes the variables are measured on continuous scales that have normal distributions. I believe the Fischer's Z transformations are based on correlations of normally distributed variables. The Spearman rho is a nonparametric statistic.
What would one have to do if I have two correlation coefficients taken from the exact same sample (repeated measures). For example, I want to compare the relationship between sleep quality and optimism for two moments of measuring, thus having an r correlation for measure 1 and then giving the same subjects the exact same questionnaires at measure 2 and obtaining a second r correlation. Now I want to compare those two r's. How do I do that? Because I have dependent samples and only 2 variables. I have almost searched the whole internet and feel totally stuck... Could you maybe help me?
Thanks for the vedio. It was useful. I was wondering if you can confirm whether the procedure you are using in this vedio to test the difference of coefficients of correlation for Pearson r is valid for correlation coefficients obtained from Spesrman rho as well???
Hi. No, I do not believe this procedure is valid for the Spearman rank-order correlations. The procedure in the video is for Pearson correlation coefficients, which assumes the variables are measured on continuous scales that have normal distributions. I believe the Fischer's Z transformations are based on correlations of normally distributed variables. The Spearman rho is a nonparametric statistic.
What would one have to do if I have two correlation coefficients taken from the exact same sample (repeated measures). For example, I want to compare the relationship between sleep quality and optimism for two moments of measuring, thus having an r correlation for measure 1 and then giving the same subjects the exact same questionnaires at measure 2 and obtaining a second r correlation. Now I want to compare those two r's. How do I do that? Because I have dependent samples and only 2 variables. I have almost searched the whole internet and feel totally stuck... Could you maybe help me?
did you ever find an answer?