We investigate the properties of K0V stars with Hipparcos parallaxes and spectral types taken from the Michigan Spectral Survey. The sample of 200 objects allows the empirical investigation of the magnitude selection (Malmquist) bias, which appears clearly present. By selecting those objects that are not affected by bias, we find a mean absolute magnitude of Mv~5.7, a downward revision from 5.9 mag. listed in Schmidt-Kaler (1982). Some objects have absolute magnitudes far brighter than Mv~5.7, and it is suggested that these objects (~20% of the total sample) are K0IV stars which may have been mis-classified as a K0V star. The presence of the Malmquist bias in even this high quality sample suggests that no sample can be expected to be bias-free.

The theoretical prediction that trigonometric parallaxes suffer from a statistical effect, has become topical again now that the results of the Hipparcos satellite have become available. This statistical effect, the so-called Lutz-Kelker bias, causes measured parallaxes to be too large. This has the implication that inferred distances, and hence inferred luminosities are too small. Published analytic calculations of the Lutz-Kelker bias indicate that the inferred luminosity of an object is, on average, 30% too small when the error in the parallax is only 17.5%. Yet, this bias has never been determined empirically. In this paper we investigate whether there is such a bias by comparing the best Hipparcos parallaxes which ground-based measurements. We find that there is indeed a large bias affecting parallaxes, with an average and scatter comparable to predictions. We propose a simple method to correct for the LK bias, and apply it successfully to a sub-sample of our stars. We then analyze the sample of 26 `best' Cepheids used by Feast & Catchpole (1997) to derive the zero-point of the fundamental mode pulsators and leads to a distance modulus to the Large Magellanic Cloud - based on Cepheid parallaxes- of 18.56 +/- 0.08, consistent with previous estimates.