GRB optical afterglow and redshift selection effects: The learning curve effect at work

We show how the observed gamma ray burst (GRB) optical afterglow (OA) and redshift distributions are changing in time from selection effects. For a subset of {\it Swift} triggered long duration bursts, we show that the mean time taken to acquire spectroscopic redshifts for a GRB OA has evolved to shorter times. We identify a strong correlation between the mean time taken to acquire a spectroscopic redshift and the measured redshift. This correlation reveals that shorter response times favour smaller redshift bursts. This is compelling evidence for a selection effect that biases longer response times with relatively brighter high redshift bursts. Conversely, for shorter response times, optically fainter bursts that are relatively closer are bright enough for spectroscopic redshifts to be acquired. This selection effect could explain why the average redshift, $<z>\approx2.8$ measured in 2005, has evolved to $<z>\approx2$, by mid 2008. Understanding these selection effects provides an important tool for separating the contributions of intrinsically faint bursts, those obscured by host galaxy dust and bursts not seen in the optical because their OAs are observed at late times. The study highlights the importance of rapid response telescopes capable of spectroscopy, and identifies a new redshift selection effect that has not been considered previously, namely the response time to measure the redshift.