“…In longitudinal studies, the number of identified trajectories usually ranges between two and five, and these trajectories differ according to the method chosen as well as to the population targeted (Blokland et al, 2005; Boers and Reinecke, 2007; Bushway et al, 2003; Dahle, 2005; Evans et al, 2016; Farrington et al, 2006; Loeber et al,1993; Moffitt, 1993; Moffitt et al, 1996; Murphy et al, 2012; Nagin and Land, 1993; Piquero, 2008; Prein and Schumann, 2003; Sampson and Laub, 2003; Stelly and Thomas, 2005; Thornberry, 2005; Walters, 2012). Typically, there is a large share of approximately 50 percent of non-delinquently involved (or at least hardly involved) individuals in these studies, which often focus on birth cohorts or the general population (see more recent publications, for example Boers et al, 2010; Gann et al, 2015; Murphy et al, 2012; Odgers et al, 2008, Piquero et al, 2010). If only studies using latent group-based trajectory models that identify unobserved heterogeneity in longitudinal datasets are taken into account, as was done in a systematic review by Jennings and Reingle (2012), a range of two to seven groups can be identified.…”