84). The instrument of non-verbal intelligence ( Kornmann and Horn, 2001) was developed as part of a educational screening/counseling battery, with items based on the Figure Reasoning Test (FRT) (25 items) and also validated within a large sample (N=4319, Cronbach׳s α=0.81). Together with gender, these measures allowed to control for and analyze possible influences of learner features on the effects of the intervention.

Moreover, School Type (ST) was included as covariate, due to the general educational level coming along with it. According to the variable Cell Cycle inhibitor plan and the quasi-experimental design described above, ANOVA and ANCOVA were applied as relevant methods (using SPSS in version 22). Motivation and achievement in physics served as dependent variables, while group membership, school type and gender served as independent variables as well as non-verbal intelligence, reading comprehension and pre-test physics achievement served as covariates. The reported measure of effect size is omega squared (ω2), i.e. the population estimate of (total) explained variance, with the usual size categorization (see Cohen,1988: small effects:

0.01<ω2<0.06; Osimertinib purchase medium effects: 0.06≤ω2<0.14; large effects: 0.14≤ω2). A 2×2-analysis of variance (ANOVA) was carried out using ‘prior achievement level in physics’, ‘non-verbal intelligence’ and ‘reading comprehension’ as dependent variables and group membership and school type as independent variables (descriptive data: see Table 4). Whereas the groups did not differ in any pre-test variables, Cytidine deaminase the factor ‘school type’ had a significant but small influence on non-verbal intelligence (F(1,

118)=5.6; p<0.05; ω2=0.04) and – much stronger – on reading comprehension (F(1, 118)=20.6; p<0.01; ω2=0.14) before the intervention. This fact was not surprising: because education level in school type 2 is generally significantly more demanding (see PISA-Konsortium Deutschland, 2008), students in this school type are strongly expected to have higher reading comprehension and non-verbal intelligence. For this reason, the covariates in question had to be taken into account. Furthermore, there was a small, but significant interaction of group membership and school type for motivation (total: F(1, 118)=6.8; p<0.05; ω2=0.05; “classroom climate” (CC): F(1, 118)=4.8; p<0.05; ω2=0.04; and “self-concept” (SC): F(1, 118)=6.3; p<0.05; ω2=0.06). In school type (ST) 1, measures of classroom climate (CC), self-concept (SC) and motivation in total were higher in the TG than in the CG. In contrast, the same measures were lower in the TG than in the CG in ST 2 (see Table 4). After treatment subject specific physics achievement was tested with the same instrument in both groups.