Research Keynote Address

Causal Generalizations: Building Connections Between Science and Policy

In science, we are often interested in knowing if an intervention or treatment "causes" an outcome to change. Teasing apart causality requires the use of research designs that have high internal validity (e.g., randomized experiments or strong quasi-experiments).

Outside of basic science, the results of these studies are often intended to inform policies and practice for individuals and organizations. This use calls to question the external validity of these designs.

In her research keynote, Tipton will reflect on her work as a statistician developing methods to improve the external validity of these high internal validity designs. This includes work on the design and analysis of individual field trials, as well as the collection of evidence across trials using meta-analysis.

Date: Thursday, November 4
Time: 12:00 p.m. to 1:00 p.m.
Location: Nebraska Union, Swanson Auditorium

The research keynote is free, open to the public and requires no registration.

Workshops

Workshop 1

The Generalizer: A Free Tool to Help You Design Sample Recruitment Plans

The Generalizer is a free web tool that can be used to design sample recruitment plans in education studies, including both K-12 and higher education. The tool walks researchers through the process of identifying a target population, considering potential treatment effect moderators and designing a stratified recruitment plan.

This workshop will provide an overview of the methods behind the Generalizer and an introduction to the tool itself. For those conducting studies in environments other than schools, there is also an R software package with more flexibility, which will also be introduced.

Workshop 2

Introduction to Methods for Large Meta-Analyses

In standard meta-analysis workshops and introductions, the focus is on methods for small meta-analyses (e.g., fewer than 20 studies). In education and the social sciences, however, meta-analyses are often much larger (e.g., more than 50 studies). These large meta-analyses bring with them new methodological problems, including dependent effect sizes and a shift in focus from an average effect size to understanding variation in effect sizes.

This workshop will provide an overview of "best practices" in these meta-analyses, as well as an introduction to the use of robust variance estimation. The workshop includes applications in R software using the clubSandwich and metafor packages.