Is the dependence of spectral index on luminosity real in optically selected AGN samples?
We critically examine the dependence of spectral index on luminosity in optically selected AGN samples. An analysis of optically selected high-redshift quasars showed an anti-correlation of $α_{OX}$, the spectral index between the rest-frame 2500 A and 2 keV, with optical luminosity (Miyaji et al. 2006). We examine this relationship by means of Monte Carlo simulations and conclude that a constant spectral index independent of optical luminosity is still consistent with this high-z sample. We further find that that contributions of large dispersions and narrow range of optical luminosity are most important for the apparent, yet artificial, $α_{OX} - l_o$ correlation reported. We also examine another, but more complete low-z optical selected AGN sub-sample from Steffen et al. (2006), and our analysis shows that a constant spectral index independent of optical luminosity is also consistent with the data. By comparing X-ray and optical luminosity functions, we find that a luminosity independent spectral index is in fact more preferred than the luminosity dependent spectral index model. We also discuss the selection effects caused by flux limits, which might systematically bias the $l_x - l_o$ relation and cause discrepancy in optically selected and X-ray selected AGN samples. To correctly establish a dependence of spectral index of AGNs on their luminosity, a larger and more complete sample is needed and consequences of luminosity dispersions and selection effects in flux limited samples must be taken into account properly.