Three essays on school organization, student performance, and long-term

• Autores: Jan-Christoph Bietenbeck
• Directores de la Tesis: David Franz Dorn (dir. tes.)
• Lectura: En la Universidad Internacional Menéndez Pelayo (UIMP) ( España ) en 2015
• Idioma: español
• Tribunal Calificador de la Tesis: Sara De la Rica Goiricelaya (presid.), Manuel Fernández Bagües (secret.), Jesús M. Carro (voc.), Stephen Gibbons (voc.), Pedro Mira (voc.)
• Programa de doctorado: Programa Oficial de Doctorado en Economía
• Calificación: Sobresaliente Cum Laude
• Materias:
  • Ciencias económicas
• Texto completo no disponible (Saber más ...)
• Resumen

  • In this nal chapter, I summarize the ndings, discuss the contribution, and describe the policy implications of each of the three preceding chapters in turn.

    5.1 Chapter 2 Motivated by the recent rise in grade retention in kindergarten and the early elementary grades in the United States, chapter 2 studies spillover eects from kindergarten repeaters on the short- and long-term school performance of their classmates. In doing so, it draws on data from Project STAR, which was a large-scale experiment that randomly assigned kindergarten students and their teachers to classes within schools. Identication of spillover eects in this chapter is based on the experimentally-induced variation in exposure to repeaters between students in dierent classes.

    The analysis in chapter 2 shows that on average, students who are exposed to repeaters score 9% of a standard deviation lower than their non-exposed schoolmates on an end-ofkindergarten math test. However, these same students are rated better by their teachers on a wide range of behaviors that reect non-cognitive skills about three years after kindergarten, when such skills are rst observed in the data. While the negative spillover eect on test scores fades out over time, the positive impact on non-cognitive skills appears to persist. In line with this result, students seem to benet from kindergarten exposure to repeaters in the very long run, as they are more likely to graduate from high school and to take a college entrance exam. Importantly, the size of these spillovers crucially depends on students’ demographic background, with students from more favorable backgrounds beneting the most from early-childhood exposure to repeaters.

    Chapter 2 also oers a possible interpretation of this striking pattern of results. In particular, it suggests that students may learn from adversity, i.e. that they become persistent and disciplined in the face of adverse learning conditions due to classroom disruption by repeaters. These improved personality traits then help students succeed in school in the long run, driving the positive impact of repeater exposure on high school graduation and college test-taking. Suggestive evidence presented in chapter 2 supports this interpretation of results, although alternative mechanisms that are also correlated with improvements in non-cognitive skills cannot be denitely ruled out by the analysis Chapter 2 makes three important contributions to the small previous literature on spillovers from repeaters as well as to the wider literature on peer eects in education.

    First, due to relying on experimentally-induced variation in the exposure to repeaters, the results in this chapter have a clear causal interpretation. This is not always the case for previous studies, which have relied on observational data instead. Second, this is one of the rst studies to estimate spillovers on non-cognitive skills. Finally, the analysis moves beyond the short-term perspective in previous studies by tracing spillovers all the way from kindergarten until the end of high school.

    In terms of policy recommendations, one implication that emerges from the results in this chapter relates to the heterogeneity of the impact of repeater exposure with regard to students’ demographic background: since students from more favorable backgrounds reap higher benets from sharing a kindergarten classroom with repeaters in the long run, repeaters should be assigned to classes containing this kind of students whenever possible.1 In contrast, answering the larger policy question whether grade retention is generally benecial or harmful is beyond the scope of this chapter. In particular, any answer to this question would have to take into account at least the following three impacts: (1) the eect of retention on the repeaters themselves (from the previous literature, this is likely negative); (2) the spillovers from repeaters on their new classmates, which I study here; and (3) the eect of “losing” a low-achieving repeater for the repeater’s classmates from the previous school year. I leave this issue as a topic for future research.2 5.2 Chapter 3 Chapter 3 studies the eects of traditional and modern teaching practices on students’ cognitive skills. It is motivated by a recent surge in policy proposals that call for an increase in modern, interactive teaching practices in order to raise students’ reasoning skills.

    This recommendation is seemingly at odds with the results from a small previous literature in economics, which shows that traditional but not modern teaching raises student performance on standardized tests. Chapter 3 sheds new light on this issue by using data from TIMSS, which contains separate measures of students’ factual knowledge, routine problem-solving skills, and reasoning skills.

    The analysis in chapter 3 shows that traditional and modern teaching practices promote very dierent cognitive skills in students. In particular, traditional teaching practices foster factual knowledge and competency in solving routine problems, whereas modern teaching practices promote reasoning skills. This pattern of impacts is found both for the 1Being in class with higher-achieving students might also benet the repeaters themselves, although this has not been examined in the literature so far.

    2Another topic for future research is the linearity of the spillover eects studied here. Because students are usually exposed to at most one repeater in my data, I cannot draw conclusions about the nature of spillovers from large numbers of repeaters in a classroom.

    United States and for a range of other developed countries. Chapter 3 is able to reconcile the positive impact of modern teaching on reasoning with its negative or zero impact on overall student performance by documenting that typically only a small fraction of the questions in standardized tests measure students’ reasoning skills.

    The chapter makes three important contributions to the small previous literature on the eects of teaching practices on student outcomes. First, it provides the rst comprehensive analysis of the impacts of traditional and modern teaching practices on student test scores for the United States. Second, it is the rst study to analyze whether traditional and modern teaching practices aect dierent cognitive skills in dierent ways, and the heterogeneity of impacts found in the results underlines the importance of doing so. Third, chapter 3 provides estimates of the impacts of traditional and modern teaching practices for a wide range of developed countries.

    The results in this chapter give rise to two novel policy implications. First, the positive impact of modern teaching practices on reasoning suggests that instructing teachers to emphasize student interactions might be a useful way to increase students’ reasoning skills. Importantly, this conclusion is in line with the ideas found in the policy proposals that motivated this study. Second, if policy makers and educators are serious about promoting reasoning skills over factual knowledge and routine problem-solving skills, standardized tests need to be adapted to give more weight to questions that measure these skills. Otherwise, teachers and schools have little incentive to employ modern teaching practices and to thus promote reasoning if their goal is to raise test scores.

    The chapter also suggests two pathways for future research. First, a potentially important outcome that this study has not been able to measure are non-cognitive skills. It seems plausible that modern teaching in particular, with its emphasis on student interactions, promotes non-cognitive skills such as the ability to work in teams. Future studies could use data containing measures of both teaching practices and non-cognitive skills to address this gap in the literature. Second, although the empirical strategy in this chapter addresses the most obvious concerns related to identication, the results might still be confounded by correlated unobserved teacher traits. Therefore, any study that is able to identify a truly exogenous change in teaching practices would have a lot of merit.

    5.3 Chapter 4 Chapter 4 studies the impact of teacher subject knowledge on student performance in 13 Sub-Saharan African countries. The empirical analysis uses novel data from a repeated international student assessment conducted in these countries, whose unique feature is the availability of teacher test scores. The analysis puts special emphasis on the role of teacher subject knowledge for explaining cross-country dierences in student performance, which many researchers have suspected to be driven by dierences in teacher quality.

    The results in this chapter show that at the country level, student performance is highly correlated with teacher subject knowledge both in levels and in changes over time.

    This correlation is conrmed in OLS regressions at the student level, in which we are able to control for a large range of student, teacher, and school characteristics. Our headline regressions, which additionally control for the inuence of any subject-invariant student traits, show that a one-standard-deviation increase in teacher subject knowledge raises student performance in the same subject by about 3% of a standard deviation on average.

    This impact is substantially larger for teachers with higher levels of education and experience as well as for teachers of the same gender as the student.

    Regarding the scientic contribution, the analysis in chapter 4 is the rst to study the impact of teacher subject knowledge across a wide range of developing countries.

    Moreover, it is one of very few studies in the literature that can convincingly address the identication concerns related to the sorting of students and teachers across schools and classrooms. Finally, we provide the rst evidence on the interaction between teacher subject knowledge and teacher education and experience in the production of student knowledge.

    The results in chapter 4 show that dierences in teacher subject knowledge can account for a signicant although modest part of the cross-country variation in student performance in Sub-Saharan Africa. Therefore, countries whose students perform relatively badly in the SACMEQ assessment can expect improvement in student test scores if they provide teacher training that aims at fostering teachers’ knowledge of the subject they teach. Future research should look at the role of dierent teacher traits, such as teachers’ non-cognitive skills, for cross-country dierences in student performance.