aes24 International Evaluation Conference

Authors: Mohib Iqbal (Department of Education), Kira Duggan (Department of Education, AU), Ben McNally (Department of Education, AU)

This presentation will showcase the use of quasi-experimental impact evaluation and the use of a relatively new data linkage capability within the Victorian public sector.
Impact evaluation provides important evidence on program effectiveness and helps to inform government investment decisions. Quasi-experimental design identifies a comparison group that is similar to the treatment group/program participants in terms of baseline or pre-intervention characteristics. Statistical methods such as propensity score matching, and regression discontinuity can create valid comparison groups with a reduced risk of bias (White & Sabarwal, 2014).

However, the implementation of this method faces significant technical, data availability, and other challenges.
The Evaluation and Program Impact (EPI) branch at the Victorian Department of Education (DE) used quasi-experimental assessment as part of six different education program evaluations spanning issues from teacher supply to support for vulnerable students. This approach was used to evaluate impact/effectiveness and the economic evaluation of interventions to measure avoided costs. The presentation will outline the process of design, methodology and implementation of quasi-experimental methods used as part of these six evaluations.

Key enablers of the use of quasi-experimental designs are data availability and expertise in undertaking advanced quantitative impact evaluations. This presentation will give an overview of the types of departmental data used (such as regularly administered student, parent/carer, teacher and school leader surveys, assessment results such as NAPLAN and administrative data) as well as the relatively new analytical capability available through linked service use data from the Victorian Social Investment Integrated Data Resource (VSIIDR) and Centre for Victorian Data Linkage (CVDL).
The presentation also contextualises quasi-experimental impact evaluations as being one component of mix-method approaches that were staged after evaluation of appropriateness, design and fidelity. Decisions on intervention effectiveness were made using a broader array of evidence including quasi-experimental impact evaluation as one of many sources.