Central Data Catalog

1991-2011 and 2011-2012:
A stratified three stage simple random sample design is used as follows:
Stage 1: Selection of schools. Strata are built from the list of schools that have students enrolled at the targeted grade. Regions or districts (this varies from one country to another), type of schools, and geographical location (urban/ rural) are used as stratification variables. In each stratum, schools are sampled using Systematic Probability Proportional to Size (PPS) sample, i.e. the number of students enrolled at grades 2 and 5. Desired sample sizes (to achieve sampling precision) for schools is 180 since 2010. Before 2010, the desired sample size was 150.
Stage 2: Selection of classroom. One classroom at the targeted grade is selected in the sampled schools using simple random sampling.
Stage 3: Selection of students. 15 students are selected at the targeted grade in the sampled classrooms, using simple random sampling. If a classroom has 15 students or less, all of them are selected. However, selected classroom must have at least 8 students, otherwise the classroom is replaced, by another classroom from the same school if there is has another classroom at the targeted grade, or by a classroom selected from the replacement schools.

From 2013 onwards:
A stratified three stage simple random sample design is used as follows:
Stage 1: Selection of schools. Strata are built from the list of schools that have students enrolled at the targeted grades. Regions or districts (this varies from one country to another), type of schools, and geographical location (urban/ rural) are used as stratification variables. In each stratum, schools are sampled using Systematic Probability Proportional to Size (PPS) sample, i.e. the number of students enrolled at grades 2 and 6. The standard sample size is 180 schools in grade 6. But if the countries want to conduct additional analysis on specific issues, the sample size may exceed 180. In grade 2, the sample size is 90 schools, half of the number of schools selected for grade 6.
The replacement school is used when the selected school refuses to participate or is located in inaccessible area. However, the replacement school is not used when a selected school does not have students at the targeted grade (Although the sampling frame should list only schools that have students at the targeted grades, this may happen if the sampling frame is out of date).
Stage 2: Selection of classroom. One classroom at the targeted grade is selected in the sampled schools using simple random sampling.
Stage 3: Selection of students. 20 students in grade 6, and 10 students in grade 2 are selected in the sampled classrooms using simple random sampling. If a classroom has 20 students or less in grade 6 and 10 students or less in grade 2, all of them are selected.

Note: The National Research Coordinator provides the sampling frame.


Between 1991 and 2012, a participation rate of 85% was the minimum requirement for schools, including replacement schools, and regardless of students' participation rate.

Since 2013, the participation rates are computed with and without replacement schools, and with and without students' participation rate.

Countries with school response rates below 85% (with replacement schools) and student response rates below 80% are discarded for the international analysis.


1991-2010 and 2011-2012:
Unweighted participation rates are computed. Multiple imputations are used to estimate missing data.

From 2013 onwards:
Weighted and unweighted participation rates are computed. A replication methodology is used to estimate the sampling variances of parameter estimates.
Any variable with 30% or more missing values is discarded from the analysis.