Guidelines for data fusion with international large scale assessments: Insights from the TALIS-PISA link database.

Highlights • We provide some guidelines for practitioners combining data from TALIS and PISA in empirical analyses. • We stress that teacherś data need to be aggregated, thus finding causal relationships is complex. • We find that students whose teachers focus on a few instructional practices obtain better results. Abstract This paper focuses on the potential uses and misuses that can be made with the so-called TALIS-PISA link created by the OECD. This is a recently developed instrument that allows for connecting data about teachers collected in TALIS with studentś data available in PISA. However, the statistical and technical aspects regarding this link between both surveys are far from straightforward. In this paper we explore the main problematic issues of the data fusion process and provide some guidelines for researchers interested in performing empirical analyses using the resulting dataset. Likewise, we illustrate some potential utilities of this dataset through the estimation of different models with the aim of exploring the relationships between teaching practices and student achievement in Spain. Our results indicate that students whose teachers concentrate on a few instructional practices obtain better results than those who have teachers using many different activities within the classroom. [ABSTRACT FROM AUTHOR]