Martine van der Pluijm

172 Chapter 6 as: ‘ Low levels of education attained refers to individuals not having attained ISCED level 3 (graded at levels 0-2), that is, not achieving beyond lower secondary education.’ (OECD, 2015, p. 15). We recommend using this narrower OECD definition of low education level and adding a category defined as attained primary school or less, as is common in Dutch policy (Roeleveld et al., 2011). A more precise definition would allow us to target specific groups of parents more accurately and acquire more refined knowledge about the effects of interventions for these target groups (Heidlage et al., 2020). The results of one of our two multiple-case studies presented in Chapter 5, show that the group of very low educated parents (primary education as highest attained level) performed fewer language activities at home at pretest, but had significantly increased these activities at posttest, in contrast to parents with a higher education (lower secondary education and higher). Future research could examine if parents in these two categories show different interactive behavior at posttest and develop differently at posttest. This requires researchers to involve enough parents from both groups of lower-education and prevent attrition. This brings us to the second recommendation. To involve substantial numbers of lower-educated parents, researchers should create a safe setting to prevent attrition during the process. Although we tried to create adequate conditions to engage lower-educated parents, our measures suffered from attrition of particularly the lowest educated parents at the posttest. An explanation for this might be that parents experienced stress during their involvement. Some parents felt uncomfortable when they were selected for observations. Teachers then explained the purpose of our research, making parents feel comfortable, and reassuring them. However, disagreement between teachers and parents also occurred occasionally, undermining feelings of trust in the aims of teachers and researchers.We recommend future researchers to create a safe environment with all participants long before the research activities begin. An optimal situation would be that some schools choose to become involved in research on a structural basis. These research locations could inform parents from the first introduction about the research that will take place at the school and the need for parental consent for research that is required when choosing this school. These research locations can normalize research and provide the opportunity to develop a general selection procedure, implying that all parents and children are involved in the research activities. This procedure prevents researchers from having to openly select a limited number of parents, who may become insecure about why they were selected, or might avoid contact with researchers. Creating a fixed number of research locations would also provide the conditions to use a randomized controlled trial with the possibilities for blind selection of participants to form an experiment and control group. A switching replications design could be employed, with all teachers and parents participating in the intervention, but creating the opportunity to compare the results of the groups with and without intervention (Trochim, Donally, & Aurora, 2014).