Martine van der Pluijm

136 Chapter 5 communicate effectively with my child”), and parent perceptions of their self-efficacy to support oral language (e.g., “ I feel successful in supporting my child’s language development”). Cronbach’s alphas for the construct (five items) are α=.73 at pretest and α=.75 at posttest. 4) Parents reported a number of language activities at home (HLE): We measured this construct on a six-point Likert scale, ranging from 1 (never) to 6 (every day). The scale Parent report of home-based involvement activities was used from the Parent Involvement Project (e.g., “I talk with my child about the school day” and “I read to my child”). We extended the scale with additional talk activities (e.g.,“I carry out language activities with my child”or“I watch television with my child”(Scheele, 2010). Cronbach’s alphas of the construct (13 items) are α=.63 at pretest and α=.72 at posttest. Procedure Pretests and postttests were conducted from October to December 2014 and from May to July 2015, respectively. In October and May, the researchers used the same procedure to recruit parents for the interviews, collaborating closely with the teachers. Teachers informed parents about the aim, the duration (30-45 minutes), and the content of the interviews. Parents could request a translator to be present. The teacher planned a date and time for the interview and informed the researcher. The interviews were conducted by a junior researcher and seven pedagogy students, who were all trained by the first author. All the students mastered one or more of the languages (Turkish, Berber, Arabic, Farsi, Urdu, and Polish) that were spoken by most of the parents and translated the questions for parents if necessary. We were not able to find translators (e.g., Bulgarian dialect, Thai) for some parents. Analyses We compared parent participation in the AHL activities at pretest and posttest. Additionally, we used repeated-measures ANOVA to analyze the development of the four constructs from pre- to post-test. Next, we used quality of delivery by teachers as a factor to explain the development from pretest to posttest (comparison of parents with teachers with a high score versus parents with teachers with a lower delivery score). Finally, we used parental education levels as a factor to explain development in the repeated measures. We checked for equality of error variances in these analyses (Levene’s test). We merged answers to the open questions and used these to interpret the results.