Sonja Mensch

87 Reliability of Movakic Table 5. Inter-rater reliability (N=19) Baseline ICC (95%CI) Mean distance of Movakic scores (SD) Total score .94** (.85-.98) 8.6 (8.7) Group of motor ability Maintaining position .89** (.35-.97) Activities .96** (.89-.98) Changing body position .85** (.57-.95) Moving around .97** (.90-.99) ICC = intraclass correlation coefficient 95%CI = 95% Confidence Interval SD = Standard Deviation *p<0.05; **p<0.01 DISCUSSION All ICC values for test-retest reliability were excellent, except for the subscale ‘Moving around’, of which the ICC was good. Here, the 95% confidence interval of ICC’s was very wide, with the lowest limit at .27. The 95% confidence interval for ‘Changing body position’ was also relatively wide compared to the other ICCs. ICC values for inter- rater reliability of the subscales ‘Maintaining position’ and ‘Changing body position’ were excellent, but the 95% confidence interval of the ICC’s was wide, with the lowest values at .57 and at .35, respectively. Adequate absolute reliability is reflected in a small difference of mean Movakic scores, for test-retest reliability 6.8 (SD 6.4) and for inter- rater reliability 8.6 (SD 8.7). Adequate reliability is also reflected in 74% and 79% of the test-retest and inter-rater distances falling within one SD from zero. However, a few respondents exhibited large distance scores, indicating suboptimal reliability for a small number of respondents making some caution in interpreting the results In fact, we expected lower ICC’s because, in accordance with the procedure, the therapists selected a relevant situation for each individual child. Their choice depended on various aspects such as diagnosis, comorbidity, therapeutic goals, used devices and abilities and disabilities. This procedure has ecological validity, i.e. represents the real-life situation, but also creates diversity, which may be reflected in the suboptimal ICC’s with wide confidence intervals. It was hard to find a more standardized procedure for this heterogeneous and specific target group. The heterogeneity of the study population is apparent fromTable 1. The children were recruited fromwide spread care organizations