37 2 research, it might be interesting to investigate the effect of factors that influence students’ scoring. In addition to this, we do not yet know the extent to which students’ Impact! scores depend on their own learning performance. If students rate their teacher’s instruction high for teaching quality, do they perform well in that subject (and the other way around)? Does that mean that the quality of teachers’ instruction is actually high, or that the students are good in that subject, for example? The research questions were answered with data from ninth graders in the context of Dutch mathematics lessons and for a specif ic timeframe during the school year (December to March). As we do not know whether the results of this study can be generalized to students in lower and higher grades, to other educational levels (e.g., to primary school), to other subjects and to other parts of the school year, it is worth answering the research questions for such other contexts as well. The results of the current study show that, with the Impact! tool, at least three measurements with at least f ive students are needed to obtain reliable scores about teaching quality. This is in line with f indings from Hill et al. (2012), who found that at least three raters and three measurement moments were suff icient to reliably measure the quality of teachers’ instruction (by means of classroom observations). From a practical point of view, it is easier to collect data from f ive students than from a whole class. However, some groups of students might then be underrepresented, such as, poorer performing students or, in case of a digital feedback system, students without a digital device. If the whole class completes the Impact! Questionnaire, information about teaching quality can be obtained that is more comprehensive. Regarding the concurrent validity of the students’ Impact! scores (whether the scores correlate with other scores; Messick, 1995), it was found that the correlation between the Impact! scores given by external raters and the ones given by students differed (Dobbelaer, 2019; Dobbelaer et al., submitted). There was no statistically signif icant correlation between the scores given by external raters and the scores given by teachers, and between the scores given by students and the ones given by the teachers. Other studies into the concurrent validity of student perceptions of teaching quality (e.g., Clausen, 2002; De Jong & Westerhof, 2001; Kunter & Baumert, 2006) showed similar results. The MET-project (2012) showed that a combination of student and observer ratings of teaching quality and previous student performance best predict future teacher performance in terms of student performance. So, both process and product/output factors are important for measuring teacher effectiveness (Cantrell & Kane, 2013). It might
RkJQdWJsaXNoZXIy MjY0ODMw