Scale Creation

• SEA-PLM test items were scaled using item response theory (IRT) scaling methodology by means of a one-parameter Rasch model for dichotomous items and the partial credit model.
• Preliminary item calibrations were conducted separately by country. To ensure consistency of item parameters across countries measuring the underlying constructs for the 3 domains, an assessment of item fit, differential item functioning (DIF) by subgroup and item-country interaction analyses were conducted.
• Final item calibrations by domain were conducted using the full country samples and then weighted by senate weights. The student-weighted likelihood estimates (WLE) and plausible values (PVs) methodology were then used on finalized regional item lists to generate values for children’s knowledge by domain. Estimations were based on the conditional item response model and the population model, which includes the regression on student background and questionnaire variables used for conditioning.
• A conditioning three-dimensional model was built for each country. The principal components were estimated for each country separately. Subsequently, the components that explained 99% of the variance in all the original variables were included as regressors in the conditioning model.
• Scale scores of all countries for each domain were normalized to 300 points and the standard deviation to 30 points, with all countries being given equal weighting. On this basis, the scores of approximately 2 in 3 children are in a range of 270 points to 330 points.
• Each proficiency scale was divided into bands describing different levels of student proficiency.
• Alignment between SEA-PLM proficiency scales and SDG indicators: The content of the final SEA-PLM proficiency bands for reading and mathematics was matched with the expanded definition of SDG 4.1.1a and SDG 4.1.1b to select the most appropriate band corresponding to the international definitions.

List of Achievement Scales

• Mathematical literacy
• Reading literacy
• Writing literacy