Page 100 - ON THE WAY TO HEALTHIER SCHOOL CANTEENS - Irma Evenhuis

Chapter 6. Process evaluation
recorded the number of invited and subscribed people and amount of reads per post. For each newsletter, statistics have been recorded of the number of people which have been: 1) sent the newsletter, 2) read it, 3) clicked on a topic to read more. As the online community and the newsletter contained several posts/newsletters, an average has been taken separately for each registered item.
Statistical analysis
School characteristics were described, and linear mixed model analyses were performed to identify differences in factors affecting implementation (dependent variable) between the intervention and control groups (independent variable). The analyses were done at both stakeholder (level 1) and school level (level 2) by including a random intercept for school in all analyses, because of the assumption that stakeholders within one school were more similar to each other, compared to stakeholders of other schools. We adjusted for the baseline measurement because of any potential differences between groups at baseline. Since the mixed model analyses revealed negligible between schools variance (threshold ICC < 0.20) [110], linear regression analyses were performed.
For the quality of implementation, mean scores were calculated for each implementation tool, per evaluation concept, and complemented by information collected by open-ended questions. These data were analysed qualitatively by hand using Microsoft Excel, by two researchers independently, following the Thematic Content Approach [97]. First, answers were labelled with objective, descriptive codes; second, codes were split, merged, and in- terpretative codes were created; third, codes were compared, correlations identified, and overarching themes were formed. Statistical analyses were performed with MLwiN version 2.36 (Centre for Multilevel Modelling, University of Bristol, Bristol, England) and IBM SPSS Statistics version 24.0 (IBM corporation (IBM Nederland), Amsterdam, The Netherlands).
RESULTS
Characteristics of the schools
Table 6.2 provides school characteristics. Most included schools already organised relevant activities (e.g., encouragement of drinking water, availability of policy, a workgroup or action plan). More intervention (n = 5) than control (n = 2) schools created a policy to restrict students to take unhealthy or big portions of food to school.
Table 6.2. Characteristics of the participating intervention and control schools.
Characteristics of the Schools
Number of students Mean (SD)
Range Educational level (n)
Only vocational
Only senior general/pre-university Vocational and senior-general/pre-university
Intervention Schools (n = 10)
928 (509)
215–1926 330–1720
33 23 5 4
Control Schools (n = 10)
1145 (503)
98
table continues