I would like to know whether it is reasonable to test the correlation between correlations in a number of samples. For example: if I wish to know whether the correlation between a and b (climate and rainfall) is similar to the correlation between a and c (climate and streamflow) at a number of locations, can I plot the correlation between those correlations? Is there a better way to assess the similarity between these relationships? (example shown in the plot below)
I might not fully understand your question. But I think you might want to check partial coherence analysis. Here is an example study by my former colleague Dr. Mohamed Siam from MIT.
Article Role of the Indian Ocean sea surface temperature in shaping ...
Correlation needs to be done separately between 2 variables, even if one variable may be common
Dear Abdou,
I'm sure that in special cases correlation plot in which both of the axes show correlations makes sence. Maybe not the correlation itself (i.e. the r and p values) but the correlation plot can help you interpreting the result. One example: if you calculate the correlation between temperature and rainfall in several years and two regions, and your abscissa is the correlation measured in Region A, your ordinate is the correlation measured in Region B, and the points in the plot are the years. Then analyzing where the points (=the years) are displayed (i.e. in which part of the plot, both horizontally and vertically, and also relative to the y=x diagonal) might assist the understanding of your data.
HTH,
Ákos
Correlation is a parameter that we can evaluate dependence between two column data like your data. you can obtain 3 output from a b c and rate them. The code at matlab is corr.