Differences in the frequencies and error estimates of p-modes between different fitters - a problem for model fitting
Ian Roxburgh (Queen Mary University of London)

Not infrequently the frequencies for a given star determined by different fitters do not agree with each other - the differences can be larger than the error estimates so that were one to find a model that fitted one frequency set it may well not fit an alternative set. I illustrate this by comparing frequency sets for 21 stars from the Kepler Legacy project with alternative frequency estimates from Davies (2015), Appourchaux (2014) and myself.
I explore the origin of the differences and compare with estimates from an independent fitting routine, concentrating on 16CygA and 16CygB. I find that there is some error in the Legacy determination of some of the frequencies of 16CygA, and that for 16CygB, whilst on using the kasoc weighted power spectrum I reproduce the Legacy frequencies, except for modes with low signal/noise, I do not reproduce these frequencies using a power spectra derived from the kasoc light curve. I find much better agreement with Davies's frequencies (except for modes with low signal/noise) on using his power spectrum, a power spectrum derived from his time series, and power spectra derived from the kasoc light curve.
Other factors that affect fitted mode parameters include different fitting techniques, different constraints on the fitting procedure, large variations in estimates at low and high frequencies where the mode power is small compared to the background, and estimates of rotational splitting. In my opinion, given that there is no known "correct" procedure, the differences between values obtained using different techniques should be included in estimates of the uncertainties in the frequencies.