Esther Mertens

118 | Chapter 6 Reliability of this first selection, based on 9% (800 studies) screened by two independent researchers, was substantial, with 98% agreement (Cohen’s κ = .71; Landis & Koch, 1977). The remaining 429 studies were studied full-text. In this step another 310 studies (72%) were excluded. Reliability of this second selection, based on 10% (45 studies) screened by two independent researchers, was substantial, with 89% agreement (Cohen’s κ = .76; Landis & Koch, 1977). Any disagreements between the researchers concerning inclusion were solved through discussion. See Figure 1 for the flow diagram. Figure 1. Flow diagram. 9498 unique records identi ed through database searching: • PsycINFO: 6102 • PubMed: 2964 • ERIC: 1683 • CENTRAL: 567 429 records screened full-text 9069 records excluded based on title and abstract 130 records relevant for study 94 records included in study 36 records requiring more data 7 records with no author response 29 records with author response Authors could not locate data ( n = 19) Information was obtained ( n = 10) 104 records included in study 22 records included from reference search: • New interventions: 15 • Related publications: 7 Full-text of 8 records not obtainable 313 records excluded: • Other aim: 53 • Not targeting students: 10 • Not universal population: 39 • Other grades: 50 • No intervention: 19 • Not school-based / not during school hours: 16 • Not group-based: 6 • No control group: 36 • No evaluation: 24 • Statistics (no premeasurement / qualitative): 13 • Newer/more comprehensive article available: 9 • Same sample and outcomes: 1 • Unpublished: 25 • Language: 12