*4.7. Statistical Analysis*

It is inappropriate to determine whether a fixed effect model or random effect model should be employed using the heterogeneity statistic I2. In order to select an appropriate effect model, the characteristics of the study, the subjects of the study, the method of intervention, and the mean value of the intervention effect were examined. In order to select an appropriate model to determine statistical heterogeneity, the characteristics of individual studies, study design, study subjects, intervention methods, and average values of intervention effects were examined [80].

Effect sizes were calculated to determine and compare the effect of AAI and PRI/different interventions on activities of daily living, stress, depression, and mental health using the sample size, mean, standard deviation, and statistically significant test of the experimental and control groups. According to the analysis criteria suggested by Cohen [81], 0.2 or less was considered a small effect size, 0.5 a medium effect size, and 0.8 or more a large effect size. The quantitative results of the meta-analysis were presented using forest plots. Publication bias was assessed by creating funnel plots. These were assessed by two reviewers and any disagreements were resolved by discussion. The chi-squared test was performed to determine the significance of the Q statistic [82,83]. If the *p*-value of Q was less than 0.10, there was deemed to be significant statistical heterogeneity between studies. A higher significance level was used since the Q statistic has low statistical power when only a small number of studies are included in a meta-analysis [84]. All statistical analyses were performed using Review Manager 5.3 software (RevMan; the Cochrane Collaboration, Oxford, UK).
