On 16/07/2012 10:10 AM, email@example.com wrote: > Hi, I have an experiment where I have a gold-standard method to calculate the concentration of a sample and two methods I want to test how well they measure the concentration and if one of them is better than the other. I have set up a concentration gradient from 0 to 5 units in breaks of 0.5 and measure each sample with each new method. > I have done an ANOVA test on the measurements obtained for each method and have found that measurements are statisticaly different between the two methods. > Then I have run a linear regression on each method to see how each method fits the gradient, and I see that one of the two methods has a higher R-squared value. > My question is, can I say based on R or R-square values that one of the methods is significantly better than the other at measuring concentration based on both the R-square value and the significance of the ANOVA test? > > Thanks, > > Dave >
The method due to Altman & Bland could be used for comparing each of the methods to the gold standard.