We present an independent confirmation of the zero-point offset of Gaia Data Release 2 parallaxes using asteroseismic data of evolved stars in the Kepler field. Using well-characterized red giant branch stars from the APOKASC-2 catalog, we identify a Gaia astrometric pseudocolor ( eff )- and Gaia G-band magnitude-dependent zero-point offset of ϖ seis − ϖ Gaia = 52.8 2.4 (rand.) 8.6 (syst.) − (150.7 22.7)( eff − 1.5) − (4.21 0.77)(G − 12.2) as, in the sense that Gaia parallaxes are too small. The offset is found in high- and low-extinction samples, as well as among both shell H-burning red giant stars and core He-burning red clump stars. We show that errors in the asteroseismic radius and temperature scales may be distinguished from errors in the Gaia parallax scale. We estimate systematic effects on the inferred global Gaia parallax offset, c, due to radius and temperature systematics, as well as choices in bolometric correction and the adopted form for Gaia parallax spatial correlations. Because of possible spatially correlated parallax errors, as discussed by the Gaia team, our Gaia parallax offset model is specific to the Kepler field, but broadly compatible with the magnitude- and color-dependent offset inferred by the Gaia team and several subsequent investigations using independent methods.