Context. Asteroseismic surface gravity values can be important for determining spectroscopic stellar parameters. The independent log (g) value from asteroseismology can be used as a fixed value in the spectroscopic analysis to reduce uncertainties because log (g) and effective temperature cannot be determined independently from spectra. Since 2012, a combined analysis of seismically and spectroscopically derived stellar properties has been ongoing for a large survey with SDSS/APOGEE and Kepler. Therefore, knowledge of any potential biases and uncertainties in asteroseismic log (g) values is now becoming important. Aims. The seismic parameter needed to derive log (g) is the frequency of maximum oscillation power (νmax). Here, we investigate the influence on the derived log (g) values of νmax derived with different methods. The large frequency separation between modes of the same degree and consecutive radial orders (Δν) is often used as an additional constraint for determining log (g). Additionally, we checked the influence of small corrections applied to Δν on the derived values of log (g). Methods. We use methods extensively described in the literature to determine νmax and Δν together with seismic scaling relations and grid-based modelling to derive log (g). Results. We find that different approaches to derive oscillation parameters give results for log (g) with small, but different, biases for red-clump and red-giant-branch stars. These biases are well within the quoted uncertainties of ~0.01 dex (cgs). Corrections suggested in the literature to the Δν scaling relation have no significant effect on log (g); however, somewhat unexpectedly, method specific solar reference values induce biases close to the uncertainties, which is not the case when canonical solar reference values are used.