“…A variety of approaches for estimation of g have been considered in the literature: a maximunl likelihood method is given in Snyder et al (1988), B-Splines are used by Mendelsohn and Rice (1982), and in Masry and Rice (1992) Gaussian deconvolution is based on estimates of derivatives of g. Liu and Taylor (1989) seems to be the first published work that investigates the performance of kerneltype estimators in this context. Other work on deconvolving kernel estimators includes Zhang (1990), Stefanski (1990), Fan et al (1990), Stefanski and Carroll (1990), Fan (1991aFan ( , 1991bFan ( , 1992, Fan and Truong (1993), Fan and Masry (1992), and Masry (1991aMasry ( , 1991bMasry ( , 1993aMasry ( , 1993b.…”