# The Colombian Escuela Nueva school model: Linking program implementation and learning outcomes

This dissertation uses a mixed methods design to analyze how the Colombian student-centered school model Escuela Nueva affects learning outcomes, and how well the model is implemented. Primary data from 78 schools in the department Quindío show large variation in implementation across schools, both overall and with regard to the model elements. On average, schools implement only around 62% of the elements. While schools that are officially classified as Escuela Nueva tend to implement more elements than conventional schools, the difference is not large, and considerable variation exists within each group. Qualitative data confirms these heterogeneities, and suggests that differences across schools are even larger than captured by the quantitative data, given the different ways in which the program is being used or adapted in practice. Learning outcomes are measured as scores on the national standardized test Pruebas SABER. Multilevel modeling techniques are used to analyze the scores from over 810,000 students in 21,235 schools across Colombia. The results show that students in schools that are officially classified as Escuela Nueva score significantly better, the difference amounting to 10.5 to 23.2 points (0.14 to 0.30 standard deviations). This effect is comparable to the effect of the difference of one socioeconomic level. Furthermore, Escuela Nueva tends to decrease the achievement gaps between socioeconomic levels and genders. The analysis also reveals large differences in the effect of the school model across municipalities and departments. For the department Quindío, the effect of the school model is analyzed using an implementation index instead of the official classifier. Data is available for 1,068 students in 76 schools, representing half of the department’s rural primary schools. Multilevel estimation generally shows no effect of program implementation, but cannot take into account the large relative sample size. Survey estimation techniques reveal a large effect of Escuela Nueva implementation for grade 3 mathematics and for civic competencies, where the difference in the expected score between a school with a low and one with a high implementation index is 140 to 220 points. The department-level analysis also confirms that the Escuela Nueva model helps to close gaps between socioeconomic levels.