Does anyone know the right formula to get t-value for the difference of EC50 of 2 drugs calculated according to the R.J.Tallarida`s method (regression analysis)? I have EC50 values, variance, standard error, df
t = the sum of the differences of EC50 values divided by the number of subjects (N) all divided by the square root of the sum of the differences squared minus the sum of the differences divided by N divided by (N - 1) x N. You don't need var., SE or df
Assuming the comparison makes sense, and that underlying assumptions are fullfilled, you could build the standard deviation of the difference (as the square root of its variance, which is itself the sum of the variances of the two EC50 assuming independant experiments), then use it to compute your t-statistic as (EC50(2) - EC50(1))/(standard deviation of the difference). Df would then be the sum of the DF of each EC50.
However, you should read about statistics, regression and so on to be sure not to overuse this way of computing.