On selection effects in pulsar searches

Abstract

Selection effects are a major source of error in statistical studies of pulsar data since the observed sample is a biased subset of the full galactic pulsar population. It is important to identify all selection effects and make a reasonable model before attempting to determine pulsar properties. Here we discuss a hitherto neglected selection effect which is a function of the period P of the pulsar. We find that short-P pulsars are more difficult to detect, particularly if their dispersion measures are high. We also discuss a declination-dependent selection effect in the I1 Molonglo Survey (I1 MS), and find some evidence for the existence of both selection effects in the pulsar data from this survey. We discuss the implications of these additional selection effects for the recently proposed 'injection' of pulsars whereby pulsars seem to switch on only at longer P. Using the I1 MS data we calculate that the observability of pulsars with P between 0.0 s and 0.5 s is about 18 per cent less with the new selection effects than hitherto believed; the mean correction is 6 per cent for P between 0.5 s and 1.0 s. We conclude that injection is not qualitatively affected by these corrections.