“…In addition to current smoking status and consumption intensity, many surveys routinely collect retrospective information on age at smoking initiation as well as former smoking status and age at (or time elapsed since) smoking cessation. We follow the substantial literature which uses such retrospective smoking data to reconstruct smoking prevalence rates and analyse smoking patterns (i) across the life-course and over successive cohorts (Kemm, 2001;Kenkel, Lillard, & Liu, 2009;Kenkel, Lillard, & Mathios, 2003b;Laaksonen et al, 1999;Lillard & Christopoulou, 2015), and (ii) with respect to educational gradients (Christopoulou, Lillard, & de la Miyar, 2013;Federico, Costa, & Kunst, 2007;Pampel, 2005;Pampel et al, 2015).…”