Innovative Statistical Techniques (and software) for the Analysis of International Large-Scale Assessments in Education

Project: Research-related funding

Description

The aim of this project is to strengthen our links with a research team based in UNAL and ICFES to develop software routines in an open source software (R) to be able to carry out specific statistical analyses with International Large-Scale Assessment (ILSA) data.

This collaboration will be based on work currently being developed by research teams in Bath and UNAL. Both research teams are currently developing software routines in R to analyse ILSA data: Quantile Regression Models in the case of Bath, and Analysis of Small Areas in the case of UNAL/ICFES.

It is important to say that currently there are no commercial software packages available to carry out these kind of analyses with ILSA data. While these two analysis techniques are not new to fields like Geography or Economics, they are not common in Education. The reason is that the complex sample and assessment design of the data requires the application of specific statistical procedures to produce robust estimates. This poses important challenges, but preliminary talks (held during the FHSS visit to UNAL) suggest that each team could benefit from the experience and expertise of the other in order to save the obstacles currently faced.

The development of software routines in R, will allow the research community to address issues that have not been well explored so far. Quantile regression can be used to answer questions like: Are the factors associated with educational outcomes the same for low, average and high performing students? The analysis of Small Areas, on the other hand, will allow us to address the issue of non-representative samples for small groups or small geographical areas. Currently, ILSA’s work with samples that are representative of whole countries, but do not represent accurately states or other smaller political demarcations. Their samples are also not designed to represent small groups like immigrants, ethnic or religious minorities. The analysis of small areas would allow us to obtain robust statistical estimates for these groups.
Short title5000
StatusFinished
Effective start/end date15/06/1731/12/17