How Cohesion Matters 4 73 Sampling and procedure Our aim was to capture ideas across the various OPPSs and not to generalize (Crouch & McKenzie, 2006). Therefore, we used a sampling strategy of maximum variation (Patton, 2002) by paying attention to the variety of the schools’ official denominational backgrounds, locations in villages, towns or large towns, and the presence of orthodox Protestant believers in the area. We did not involve more than one school in the same location. For the sake of practicality, all schools were located within a radius of 50 kilometers around Utrecht city.37 In this area, both Bible belt and urban regions were involved. Once we had an overview of all schools that could be involved in our categories, we made a random selection and contacted a school in each category. When there was no teacher available to participate, we contacted the next school in the same category. In total, we approached 39 schools (with a 38.5% positive response). Each participant was interviewed twice by the first author, with a maximum of 65 days between the two interviews. These interviews took place at the participants’ schools.38 The interviewer was familiar with the participants’ religious backgrounds because she lived in the neighborhood of these orthodox Protestant communities and worked at an orthodox Protestant teacher training institute. This appeared to be especially helpful for gaining the participants’ trust, as well as for understanding the participants’ religious language. However, there was still distance between the researcher and the participants, because the researcher was not fully part of the stricter orthodox Protestant subculture. For example, she did not attend an OPPS herself as a child. That made it easier to maintain a critical stance. Participants received a small present for their participation, and school teams were invited to freely attend a presentation of the results of the research. We followed the ethical guidelines and the guidelines of data management for scientific research at Dutch universities (DANS, 2015; VSNU, 2005, 2012). Materials The interviews were semistructured with an interview guide based on an earlier theoretical study (Markus et al., 2021a) and interview questions from earlier research by 37 In this area, approximately 52% of all Dutch “reformatorische scholen (schools)” can be found and 29% of all “gereformeerd vrijgemaakte scholen (schools)” (Dienst Uitvoering Onderwijs, 2016). See footnote 1. 38 Because of the teachers’ preferences, one situation differed from the others in several respects: in this situation, two colleagues of the same school were interviewed together during one interview lasting 160 min (instead of one teacher being interviewed twice), and this interview was located at the office of the interviewer (instead of the participants’ school). One of the other interviews was at the interviewer’s office as well for practical reasons.