Jasper Faber

Chapter 2 30 In phase three, we imported the Likert scores of the concepts and categories obtained from the questionnaire as variables into SPSS. We performed k-means cluster analyses on the concepts based on Euclidian distance for health, healthcare, and eHealth with SPSS. We determined the optimal number of clusters with the Elbow method using the factoextra and NbClust packages in R. We used an ANOVA to identify the concepts with significant (p < 0,05) contribution to the cluster segmentation. The concepts with an insignificant contribution were removed from further analysis. To validate the clusters, we performed an ANOVA with the category scores as independent and the clusters themselves as dependent variables. Using a post-hoc ANOVA, we defined the resulting clusters based on significant differences between mean scores of the concept variables. We created profiles by further clarifying and enriching these clusters by analyzing the qualitative data from the questionnaire and focus group discussions. This was done by extending on the existing categories and concepts and using the same grounded-theory approach as used in previous phases. Finally, we performed a principal component analysis (PCA) using the factoextra package in R to discover correlations between concepts from different profiles. 2.2.4 Ethics The study protocol was approved by the Human Research Ethics Committee of Delft University of Technology (approval numbers 953, 1064, and 1141). Through our relationship-based CBPR approach we aimed to limit the impact of emotional and ethical challenges such as perceived harm, feelings of stigmatization, and anxiety toward research and the research team. In the first phase, we briefed our participants orally about the nature of the study as a formal written consent in this first introduction phase would obstruct a trustful interaction. The participants provided their consent verbally to the researcher (JF). In phases 2 and 3, when the relationship was more solid, written informed consent was provided. 2.3 Results 2.3.1 Participants During the unstructured interviews in the first phase, we spoke with 16 different members of the community center. These members consisted of eight Vi’s, two Vo’s and six Kp’s. In the second phase, we interviewed five Vo’s and five Vi’s. In phase one and two, we did not collect demographic data. In the third phase, 45 participants responded to the questionnaire. From these latter responses, we excluded three participants not living in

RkJQdWJsaXNoZXIy MTk4NDMw