In the educational literature on peer effects, attention has been brought to the fact that the mechanisms creating peer effects are still to a large extent hidden in obscurity. The hypothesis in this study is that the Frame Factor Theory can be used to explain these mechanisms. At heart of the theory is the concept of “time needed” for students to learn a certain curricula unit. The relations between class-aggregated time needed and the actual time available, steers and hinders the actions possible for the teacher. Further, the theory predicts that the timing and pacing of the teachers’ instruction is governed by a “steering criterion group” (SCG), namely the pupils in the 10th-25th percentile of the aptitude distribution in class. The class composition hereby set the possibilities and limitations for instruction, creating peer effects on individual outcomes. To test if the theory can be applied to the issue of peer effects, the study employs multilevel structural equation modelling (M-SEM) on Swedish TIMSS 2015-data (Trends in International Mathematics and Science Study; students N=3761, teachers N=179). Using confirmatory factor analysis (CFA) in the SEM-framework, latent variables are specified according to the theory, such as “limitations of instruction” from TIMSS survey items. The results indicate a good model fit to data of the measurement model. The SEM-model verify a strong relation between the mean level of the SCG and the latent variable of limitations on instruction, a variable which in turn has a great impact on individual students’ test results. Thus, the analysis indicates a confirmation of the predictions derived from the frame factor theory and reveals that one of the important mechanisms creating peer effects in student outcomes is the effect the class composition has upon the teachers’ instruction in class.

With educational effectiveness studies, school-level aggregates of students' characteristics (e.g. achievement) are often used to assess the impact of school composition on students' outcomes – school compositional effects. Empirical findings on the magnitude and direction of school compositional effects have not been consistent. Relevant methodological studies raise the issue of under-specification at level 1 in compositional models - evident when the student-level indicator on which the aggregation is based is mis-measured. This phenomenon has been shown to bias compositional effect estimates, leading to misleading effects of the aggregated variables – phantom compositional effects. My thesis, consisted of three separate studies, presents an advanced methodological framework that can be used to investigate the effect of school composition net of measurement error bias. In Study 1, I quantify the impact of failing to account for measurement error on school compositional effects as used in value added models of educational effectiveness to explain relative school effects. Building on previous studies, multilevel structural equation models are incorporated to control for measurement error and/or sampling error. Study 1a, a large sample of English primary students in years one and four (9,059 students from 593 schools) reveals a small, significant and negative compositional effect on students' subsequent mathematics achievement that becomes more negative after controlling for measurement error. Study 1b, a large study of Cyprus primary students in year four (1694 students in 59 schools) shows a small, positive but statistically significant effect that becomes non-significant after controlling for measurement error. Further analyses with the English data (Study 2), demonstrates a negative compositional effect of school average mathematics achievement on subsequent mathematics self-concept – a Big Fish Little Pond Effect (BFLPE). Adjustments for measurement and sampling error result in more negative BFLPEs. The originality of Study 2 lies in verifying BFLPEs for students as young as five to eight/nine years old. Bridging the findings related to students' mathematics self-concept (Study 2) and the findings on students’ mathematics achievements (Study 1a), I demonstrate that the prevalence of BFLPEs with the English data partly explains the negative compositional effect of school average mathematics achievement on students' subsequent mathematics achievement. Lastly, in Study 3 I consider an alternative approach to school accountability to conventional value added models, namely the Regression Discontinuity approach. Specifically, I use the English TIMSS 1995 primary (years four and five) and secondary (years eight and nine) data to investigate the effect of one extra year of schooling on students' mathematics achievement and the variability across schools in their absolute effects. The extent to which school composition, as given by school average achievement, correlates with schools' added-year effects is addressed. Importantly the robustness of the RD estimates to measurement error bias is demonstrated. My findings have important methodological, substantive and theoretical implications for on-going debates on the school compositional effects on students' outcomes, because nearly all previous research has been based on traditional approaches to multilevel models, which are positively biased due to the failure to control for measurement error.

In der vorliegenden Studie wurde sowohl der Einfluss des Schulanteils an Kindern mit einer anderen Muttersprache als Deutsch auf die Sprachleistung (Leseverständnis, Wortschatz und Rechtschreibung) deutschsprachiger Kinder als auch jener des Anteils deutschsprachiger Kinder auf die Sprachleistung von Kindern mit Migrationshintergrund untersucht. Es nahmen 211 Kinder aus der vierten Grundschulstufe, 84 mit Deutsch als Muttersprache (39.8%) und 127 mit einer anderen Muttersprache als Deutsch (60.2%), teil. Kognitive Grundfähigkeiten und der sozioökonomische Hintergrund wurden miterhoben. Die mehrebenenanalytische Auswertung zeigte Effekte der kognitiven Grundfähigkeiten und des sozioökonomischen Hintergrundes auf die Sprachleistung. Die ethnische Zusammensetzung der Schule war nicht bedeutsam. Bei Kindern mit Migrationshintergrund erwies sich der Zeitpunkt des Spracherwerbs als wichtig für die Sprachleistung. Ein möglichst frühzeitiger Erwerb der Sprache des Einwanderungslandes ist daher dringend zu empfehlen. (Autor/inn/enref.) / Series: Discussion Paper Series / Center for Empirical Research Methods