Oscillation frequencies of even the best stellar models differ from those of the stars they represent, and the difference is predominantly a function of frequency. This difference is caused by limitations of modeling the near-surface layers of a star. This frequency-dependent frequency error, usually referred to as the "surface term" can result in erroneous interpretation of the oscillation frequencies unless treated properly. Several techniques have been developed to minimize the effect of the surface term; it is either subtracted out, or frequency combinations insensitive to the surface terms are used, or the asteroseismic phase $\epsilon$ is used to determine a match between models and stars. In this paper we show that no matter what technique is used to account for the surface term, as long as the physics of the models is the same, the global parameters of a star --- mass, radius and, age --- obtained from frequency analyses are robust. This implies that even though we cannot model the internal structure of stars perfectly, we can have confidence in all results that use stellar global properties obtained through the analysis of stellar oscillation frequencies.