SREE 2017 Workshop on Endogenous Subgroup Analysis Using ASPES
This Society for Research on Educational Effectiveness (SREE) workshop briefly introduces a range of methods (structural equation modeling, instrumental variables, propensity score matching, and principal stratification) used to estimate impacts on endogenous subgroups in order to situate the ASPES method among these options. The workshop also demonstrates how to use ASPES in practice, offers an applied example, and focuses on what research directors and analysts need to know in order to use ASPES. Additional topics discussed include ASPES data and sample size requirements and how to define the endogenous variable of interest. Laura Peck, Eleanor Harvill, and Shawn Moulton of Abt Associates presented the workshop.
Workshop Slides
Recorded Presentation
The Social Policy Impact Pathfinder (SPI-Path) Individual User Guide
The Social Policy Impact Pathfinder (SPI-Path) Individual User Guide (Moulton, Peck, & Bell, 2014) provides a detailed description of the analytic steps and decisions involved in using the discrete ASPES method. The guide presents the discrete ASPES method in detail, provides practical guidance and examples from the literature, and includes SAS and Stata code templates analysts can use to conduct each stage of the discrete ASPES method.
SPI-Path|Individual User-Guide
Continuous ASPES SAS and Stata code templates
The SPI-Path|Individual User-Guide provides SAS and Stata code templates for the discrete ASPES method. Below we provide SAS and Stata code templates for researchers interested in applying the continuous version of the ASPES method. We also provide a mock dataset researchers can use to test the ASPES code.
SAS code template
Stata code template
Mock Dataset
How Does ASPES Compare to Other Mediation Analysis Methods?
We provide a table which compares ASPES to alternative methods used for mediation analysis, including instrumental variables and structural equation modeling. The methods are compared in terms of the research questions addressed, estimation process, interpretation of estimates, and key assumptions.