Elien Neimeijer

31 the other hand, items with an ICC that approximates 1 indicate that variation is mainly between groups, instead of within clients. Since our goal was to measure group climate, a group construct, it is important to examine the variance of scores at the between-group level. Results Results for the Group Climate Instrument indicated a good fit to the data. Confirmatory factor analysis was conducted on all 29 GCI items. Results showed factor loadings rang- ing from .234 to .828 (Table 1). The model showed an acceptable fit to the data: χ2(334) = 457.152 (p < .001); CFI = .931; TLI = .922; RMSEA = .048 (90% CI = .036 - .058); SRMR = .071. The ratio between the χ2 statistic and the degrees of freedom was 1.37. One item of the repression subscale (i.e., ‘Clients must ask permission for everything.’ ) did not load sig- nificantly on the repression factor, and was deleted from the model to improve model fit. Further analyses were conducted with 28 items. Table 1 presents the final factor solution, showing the items and the corresponding factor loadings. The model that best fitted the data contained four first-order factors: support (11 items), growth (6 items), group at- mosphere (6 items), and repression (5 items), and a second-order factor ‘overall climate’.