73 Randomized controlled trial of a social network intervention 3 Assessment of Quality of Life (MANSA) (Priebe, Huxley, Knight, & Evans, 1999). Exclusion criteria are: (1) acute psychotic symptoms according to the clinician or DSM-IV-TR/5 criteria as measured with the MINI International Neuropsychiatric Interview-plus (version 5.0; MINI-plus) (Sheehan et al., 1998), (2) current high risk for suicide requiring immediate intervention according to the clinician and/or DSM-IV-TR/5 criteria as measured with the MINI-plus, (3) severe addiction problems as measured with the Health of the Nations Outcome Scales (HoNOS) (Wing et al., 1998) requiring immediate intervention or hospitalization, (4) a current high risk for severe aggression towards clinicians or others as measured with the HoNOS, and (5) patient is included in another scientific research project conducted at Inforsa. Sample size As our primary outcome variable is the total mean score on a self-report questionnaire – Mental Health Continuum-Short Form (MHC-SF) (Keyes, 2002), which measures mental wellbeing using Likert scaling, we consider this outcome as a continuous variable. Normality of the distribution will be checked after data collection using optimal scaling or scatter plotting. When the assumption of normality is violated, resampling methods will be used to transform the data into a normal distribution. An accurate sample size estimation is challenging since no other studies have been performed with a comparative experimental design and intervention within the same population. Former studies evaluating mentoring programs show small to medium effect sizes (Cohen, 1977, 1992; Lipsey, 1990; Rhodes, Reddy, Grossman, & Maxine Lee, 2002), whereas studies investigating psychological, educational, and behavioral treatments show larger mean effect sizes (Lipsey & Wilson, 1993). Effect sizes of life coaching performed in a nonclinical sample show medium to large effect sizes (Grant, 2003). Given the foregoing, we expect a small to medium improvement of mental wellbeing in the FNC condition compared to TAU. We consider this a realistic and clinically meaningful effect. Our first power calculations (conducted with G*Power) indicated that 150 participants are needed (75 per arm) to detect a small-to-medium effect size (Cohen’s f = .20), given an alpha of 0.05 (two-tailed) and a power of 0.80. Since this power calculation did not account for the repeated measures design of the study, more modern power calculations were performed (conducted with Stata) to calculate the required sample size. To detect a small-to-medium effect size (Cohen’s f = 0.20) in a pairwise comparison of pre-post change between two active treatment arms, with a power of 0.80 and a within-person correlation coefficient of 0.50, 68 participants are needed in this study (34 per arm). We expect 30% of the participants to be early dropouts from the study (Cullen, Soria, Clarke, Dean, & Fahy, 2011), therefore a total of 105 participants will be included in the study.
RkJQdWJsaXNoZXIy MjY0ODMw