I am currently testing the independence of two unpaired conditions A and B; with each condition representing a set of 12 values (aka bins) that compose a one-phase association exponential curve. Each bin represents a probability that falls within a specific interval (e.g. bin 1 = 0-7.5, bin 2 = 7.5-12.5). Note that the bin intervals are not equally ranged to each other. The general trend is that: with increasing bins (and by proxy, increasing intervals), the probabilities increase.
Sign test resulted with 0.0032, suggesting that since p<0.05, the conditions are significantly different with 95% confidence.
However, the CMH Chi Square resulted with p=0.168, suggesting that these conditions are not significantly different.
I am wondering is there another test that could measure the magnitude of variation/similarity between the curves?