Several effects may bias samples of blazars, causing them to miss objects which fall within their survey area and flux limits. Many, but not all, are intimately tied up with the question of classification (§ 3). We will attempt to discuss each of these effects in turn. They include the the Browne & Marchã (1993; BM) effect, lack of consistent (or consistently applied) identification criteria in some samples, and the effect of continuum variability on spectral indices. Tied up with the second topic is the question of whether all broad-line, flat-spectrum objects should be classified as FSRQs, or whether objects in which the continua are dominated by galactic light should be referred to as BLRG.
It is safe to say that there is probably no survey which is completely immune from selection effects. The impact of selection effects upon most previous surveys is not known, but must be understood to make progress towards better understanding the AGN phenomenon. The BM (1993) effect has caused Perlman et al. (1996b) and Stocke et al. (1997) to reconsider the makeup of the original Morris et al. (1991) complete sample of EMSS BL Lacs, adding three BL Lacs to the C-EMSS sample after an exhaustive perusal of the available X-ray, radio and optical data, followed by further ROSAT HRI observations and optical spectroscopy. These two papers represent the only serious attempts to reformulate existing samples of blazars to minimize selection effects. Other existing samples should be re-considered in similar fashion.