Martine Kruijtbosch

76 Chap t e r 3.1 Data collection and data analysis This cross-sectional study used the PEP-NL test with Dutch community pharmacists. These pharmacists were either early career pharmacists who completed the PEP-NL test as an assignment at the start of classes on professionalism and pharmaceutical ethics in their postgraduate education or were supervisors of early career pharmacists who completed the test at the start of a course on pharmaceutical ethics. WG distributed and collected the assignments. There were several control questions included among the 12 statements in order to correct for respondents providing socially desirable answers. 15 If respondents ranked such a control question more than one time, their tests were excluded from the study. First, a principal component analysis (PCA) was performed to check the PEP-NL rating scores for construct validity. The PEP-NL rating scores were checked for factorability with the Kaiser-Meyer-Olkin’s measure (KMO). This measure should ideally be over 0.6. Subsequently, correlations between variables were tested with Bartlett’s test of sphericity (index p < 0.05). Varimax rotation was used to extract the components to increase interpretability. The components were examined by their percentage of variance explained, their eigenvalues (eligible value > 1) and their component statement loadings (eligible value ≥ 0.35). 36 If statements loaded highly on more than one component (cut-off less than 0.2 difference between components), these were excluded. Second, Cronbach’s alpha was used to investigate the internal reliability of the remaining eligible statements of each component and the test as a whole. A Cronbach’s alpha equal to or greater than 0.70 was considered reliable. Both the PCA and Cronbach’s alpha calculations were performed using SPSS version 23. Third, the eligible statements of each component were compared with the eligible statements of each component of the PCA performed in the PEP study 33 and checked against the moral schemas of the PEP test. In case of differences within the clustered statements per component, three members of the research group (MK, AF and MB) examined these statements and labelled, through consensus, a possible new moral reasoning schema. Final consensus on the moral reasoning schemas was reached after a consulting meeting with an expert panel of five senior pharmacists and MK, WG, AF and MB.