Saskia Baltrusch

161 Chapter 6 2.4 Testing Procedure First, participants had to fill in the questionnaire related to their history of low- back pain. These outcomes were used to allocate the participants into healthy and low-back pain groups for sub-group analysis. Each measurement session began with fitting and adjusting the exoskeleton to the participant. Before the start of the measurement, every participant was allowed to move in the exoskeleton for a couple of minutes, to get familiarized with it. Subsequently, participants performed the functional performance test battery in two conditions: 1) with the exoskeleton (Exoskeleton condition) and 2) without the exoskeleton (Control condition). The support of the exoskeleton was turned on with the manual clutch for the task group 1 (tasks that are expected to benefit from the exoskeleton support), for the other tasks the support was turned off. Engagement of the clutch was set at 0 degrees. The starting condition and the sequence of the tasks were randomized to prevent order and habituation effects, but was the same for both conditions. After each separate task, participants were asked to fill in the VAS scales to assess subjective performance. At the end of the session the User’s Impression questionnaire was completed by the participants. For each participant, all measurements were performed on a single day. Half of the measurements were performed in a mock-up environment at Mitsubishi Turbocharger and Engine Europe, the other half in a mock-up environment at the VU Amsterdam. 2.5 Data Analysis We conducted two-way repeated-measures ANOVAs, to test for the effect of exoskeleton use (2 levels) and the effect of allocated group (2 levels) on mean objective performance. In case of an interaction effect, Bonferroni post-hoc tests were conducted to determine differences between exoskeleton conditions in the two groups separately. To test for statistical differences in subjective performance, the VAS scale values of perceived task difficulty and local discomfort were compared between the conditions, using the non-parametric Wilcoxon test, since VAS scales generate ordinal data [18]. For sub-group analysis of subjective data, we performed the non-parametric Mann-Whitney test to determine differences between exoskeleton conditions for each of the groups separately. The critical level of significance was an alpha of 0.05. Corrections 6