54 Chapter 2 2020) used cluster randomisation, and Poppelaars, Lichtwarck-Aschoff, et al. (2018) and Russell and Newton (2008) randomly allocated individuals in pairs to the conditions. All other studies used a regular RCT design. See Table A.11 (Appendix) for the characteristics and findings of these studies. Effect sizes were calculated for positive affect/emotions and negative affect/emotions separately. Effect sizes for positive affect/emotions were calculated on data from three papers, with self-reported positive affect (Poppelaars, Lichtwarck-Aschoff, et al., 2018; Russell & Newton, 2008) or vigour (Andrade et al., 2019) as the chosen outcome measures. For four studies data could not be obtained (Alloway & Carpenter, 2021; Branton et al., 2014; Douris et al., 2012; Gheller et al., 2019) and one paper was not included in the forest plot as it used the same data as an already included paper (Andrade et al., 2020). Regarding the papers included in the forest plot, Andrade et al. (2019) compared an active condition to casual (exer)games. For Poppelaars, Lichtwarck-Aschoff, et al. (2018), the condition in which participants received a mental-health focused introduction message was coded as active condition and the condition in which participants received an entertainment-focused introduction message was coded as casual game. In Russell and Newton (2008), three intervention conditions were employed: the conditions of regular bicycle ergometer exercise, interactive video game bicycle ergometer exercise and video game-only were coded as active condition, casual game+ and casual game, respectively, contributing two effect sizes to the plot. As can be seen in Figure A.12 (Appendix), the three effect sizes comparing an active condition to casual game(s) ranged from –0.06 to –0.50. Only Andrade et al. (2019) showed a significant effect favouring the casual exergames. The comparison of the interactive video game bicycle ergometer exercise (casual game+) with the video game-only condition (casual game) showed a nonsignificant effect size of 0.24. For negative affect/emotions, effect sizes were calculated on data from nine papers. Outcome measures used were self-reported negative affect (Matheson et al., 2021; Poppelaars, Lichtwarck-Aschoff, et al., 2018; Russell & Newton, 2008), anger (Andrade et al., 2019), hostile feelings/state hostility (Ferguson & Rueda, 2010; Ferguson et al., 2016; Valadez & Ferguson, 2012), and current stress (Hunter et al., 2019), or heartrate in response to a mental arithmetic task (Goodie & Larkin, 2001). Data from five studies were not included in the plot, as data could not be obtained (Alloway & Carpenter, 2021; Branton et al., 2014; Douris et al., 2012; Gheller et al., 2019) or the paper used the same data as an already included paper (Andrade et al., 2020). Regarding the specific intervention arms, the same notes as outlined above apply to
RkJQdWJsaXNoZXIy MTk4NDMw