Interpreting CU Triaxial results for effective parameters

I recently had a CU triaxial series run on a sample of lean clay with pore pressure monitoring and back pressure saturation. When I look at the results, they varied significantly depending on how the results were presented. Three tests were run for the series, at confining pressures of 1/2, 1 and 2 tsf. When the failure criterion was "maximum stress ratio", the effective friction angle was 32.6 degrees. When "maximum pore pressure" was used, it was 31.1 degrees. When "percent axial strain" was used, it was 31.8 degrees, but when "maximum deviator stress" was used it was only 18.9 degrees. Now I'm not sure which values would be appropriate for the effective stress analysis. Also, what could cause such a variation? Thanks