“…Nine of the studies (60%) used randomized designs with the unit of randomization being at the individual level for five of the studies (Barnes, Treiber, & Davis, 2001;Davey & Neff, 2001;Hains & Szyjakowski, 1990;Weigensberg et al, 2009;Wright, Gregoski, Tingen, Barnes, & Treiber, 2011) and the group level (e.g., school, classroom) for four studies (Barnes, Treiber, & Johnson, 2004;Jellesma & Cornelis, 2012;White, 2012;Yahav & Cohen, 2008). Those that randomized at the group level had two of five schools or classrooms in the study to randomly assign to intervention or control conditions, which was not adequate to ensure confounders would be equally distributed across the sample (Barnes et al, 2004;Jellesma & Cornelis, 2012;White, 2012;Yahav & Cohen, 2008).…”