Behavioural control was measured using three questions on control over implementation, resources and budgets. For the total scale, the mean of the three questions was calculated. The Cronbach alpha for the scale was 0.87.
The subjective norm was measured by two constructs: normative beliefs (the perception of the (dis)approval of a reference group concerning HP, e.g. how likely is it that your colleagues believe that you must invest resources in HP?) and the motivation to comply (the importance of this reference group for the respondent, e.g. how much do you care whether your colleagues approve that you invest in HP?). Normative beliefs and motivation to comply were assessed concerning eight reference-groups: the seven reference-groups from the questionnaire of Downey and Sharp [21], plus the reference-group ‘trade union’ (Table 1). As with the attitude scale, the scores of the normative beliefs and their accompanying motivation to comply were multiplied and divided by 5. The higher the score on the subjective norm, the more likely the respondent believes that HP should be implemented according to a relevant reference-group.’ A total score was calculated by taking the mean. The Cronbach alpha for this scale was 0.77.
Moral responsibility was measured using a scale developed by Hart [22]. Three dimensions of moral responsibility were included (see Table 1 for the themes). The mean was used as total score on moral responsibility. The Cronbach alpha for this scale was 0.60.
Data analysis
To investigate the current status of the implementation of HP interventions (aim 1), percentages were given of the currently implemented HP themes and actions. Chi2-test were used to analyze if these results differed by the characteristics of the workshop.
To investigate which characteristics of the workshop and individual factors of the respondent were related to the implementation of environmental HP (aim 2) and to being supportive towards implementing HP in the future (aim 3), univariate logistic regressions were used. A multivariate logistic regression was performed with all significant factors from the univariate analyses.
SPSS 21 was used to analyze the data. P-values lower than 0.05 were considered to be statistically significant.
Results