A vanishing rural school effect? Urban/rural student achievement differences in Latin America and the Caribbean

• In Event: Youth and immigration: Evaluating the effectiveness of intervention programs

Abstract

Objective
In 1997, a cross-national assessment of educational achievement in Latin America and the Caribbean found that rural schools in Colombia outperformed urban schools in tests of reading and mathematics, except in very large cities. In the 2006 follow-up to this assessment, Colombia’s rural school effect was reversed. The objective of this study is to determine whether, after adjusting for relevant background factors and applying a propensity score analysis, Colombia’s rural school advantage remained in 2006. We also seek to identify other countries in Latin America with a rural school advantage, all else equal.

Background/Conceptual Framework
In 1997, UNESCO’s Latin American Laboratory for the Assessment of Quality in Education (LLECE) undertook a comprehensive cross-national assessment of primary school achievement in 13 education systems across Latin America and the Caribbean. This study found that rural schools in the region consistently underperformed schools in smaller urban areas and in megacities, with one exception: Colombia. In third grade language tests, Colombian children in rural schools outscored those in urban schools, with the exception of children in very large cities. In tests of third and fourth grade math, rural children in Colombia outperformed children in smaller urban areas and in large cities (Casassus et al., 2002). Given vast urban/rural inequalities in most Latin American societies, Colombia’s rural school advantage merits serious attention. As Phillip Foster (1977) argued, regional and urban/rural inequalities are so great in many developing countries that they may be a more important source of inequality than differences in student socioeconomic status, race, or ethnicity.

Why did rural schools in Colombia outperform urban schools? A LLECE report published in 2002 attributes Colombia’s rural school advantage to the well-known Escuela Nueva rural school program, which was designed to improve access and quality in Colombia’s most challenging educational environments: remote, sparsely populated rural areas with insufficient numbers of students to allocate one teacher to each grade.

Analytical Methods
We used the following equation to model the relationship between student achievement and student-, teacher-, and school-level variables:

Y_i=β_0+β_1 (Student Characteristics)+β_2 (Teacher Characteristics)+〖β_3 (Rural School)+β_(4 ) (School Characteristics)+β〗_4 (pscore)+ε_i
Where Y is the outcome variable of student achievement for student i. Pscore is the propensity score included to adjust for selection bias (full paper contains detailed information). We use clustered robust standard errors to account for the nested structure and intraclass correlation of students within schools. All estimates use appropriate student weights to account for the oversampling of students during stratified random sampling. Models were run individually by country. We ran each model separately for third and sixth grade students, as well as a pooled model with all students.

Data
We used data from the Second Regional Comparative and Explanatory Study of Student Achievement in Latin America and the Caribbean (SERCE), which was conducted in 2005-2006 in 16 Latin American and Caribbean countries, as well as the Mexican state of Nuevo Leon. In each country about 5,000 children were tested in reading and mathematics. In addition to student content knowledge, SERCE collected background information from students, their families, teachers, principals, and schools.

Results
We find that rural schools in Colombia retained their advantage in only a limited way: all else equal, rural schools outperformed marginally urban schools in mathematics. However, strict comparisons between the 1997 and 2006 assessments are not possible due to differences in variables, samples, and data collection procedures. We also find that although urban schools had higher raw test scores than rural schools in most countries participating in the 2006 assessment, rural schools in several countries outperformed urban schools in adjusted scores of math and reading.

Significance
Our results suggest that all else equal, rural schools in many countries are not inherently worse than urban schools; rather, the conditions in which rural students live and study place a drag on their test score performance. If school resources like infrastructure or books strongly influence urban/rural achievement differences, then investments in such resources could be an important step toward closing urban/rural achievement gaps. In contrast, if these differences are driven largely by differences in the socioeconomic background of students and their peers, policy makers may consider more direct investments in children in rural areas, through the use of conditional cash transfers and other means to raise rural children and their families out of poverty.

Authors

• Thomas Luschei, Claremont Graduate University

• Loris Fagioli, Claremont Graduate University