Calendar

Spyros Konstantopoulos
College of Education
Michigan State University

The purpose of this study is toexamine the effects of class size on eighth grade students’ cognitive andnon-cognitive outcomes in four European countries (Hungary, Lithuania, Romaniaand Slovenia). We used TIMSS data from 2003, 2007, and 2011. Statistical analyseswere based on instrumental variable (IV) methods and a regression discontinuitydesign (RDD). Cognitive outcomes were mathematics, physics, biology, chemistryand earth science. Non-cognitive outcomes were students’ self-reports on five variablesthat were available across all three cycles of TIMSS: a) learning the subjectwell; b) learning the subject quickly; c) enjoy learning the subject; d) thesubject is hard; and e) the subject is one’s weakness. All non-cognitive items followeda Likert-type format (1=strongly agree; 2=agree; 3=disagree; and 4=stronglydisagree). The main independent variable was class size. Student,teacher/classroom and school covariates were also included in the analyses. IVestimates indicated that in Romania in 2003 smaller classes had significant andpositive effects on academic scores in mathematics, physics, chemistry andearth science and in 2007 on enjoying learning mathematics. In Lithuania in2011 smaller classes had significant and positive effects on enjoying learning biologyand chemistry and learning chemistry well. Estimates of analyses that used aRDD indicated small class effects on some non-cognitive outcomes in Lithuaniaand Romania, but only in 2007. Overall, class size effects were not systematicacross years, countries, and outcomes. The effects that were statisticallysignificant however, were also substantial in magnitude and typically muchlarger than the effects reported from Project STAR.  

Dr. Larry Hedges
Department of Statistics
Northwestern University