The logarithmic temperature dependence of resistivity, commonly observed in disordered films, has generally been interpreted as evidence for electron weak localization, with its slope indicative of the inelastic scattering mechanism. In this work, we show that the 2D quantum percolation (QP) model, pertaining to disordered metallic films, predicts a sample-size dependent ln L conductance correction that is three times larger than that for the Anderson model. Moreover, when the film has a finite thickness, the coefficient of ln L decreases to about 2 3 of its 2D value for both the QP and the Anderson models. Since for disordered metallic films the QP model is more realistic than the Anderson model, which pertains to doped metallic films, it follows that many prior experimental results on metallic films have to be re-interpreted in regard to their implications about the inelastic scattering mechanism(s). These results have direct implications for the interpretation of experimental data.