A seminar by Associate Professor Sama Low Choy from QUT
Title: Not your everyday meta-analysis—Venturing into mixed methods, machine learning, or expert elicitation
Abstract:
Meta-analysis has risen in popularity, especially since the global pandemic, when other primary methods of data collection became problematic. However, many researchers may only be familiar with the basic “everyday” version, involving systematic literature review to collate effect sizes across studies, followed by a one-way fixed or random effects analysis to pool them. Here I present three studies where we have expanded methods beyond an “everyday” meta-analysis, involving both statistical and non-statistical methods.
- In the mixed methods realm, we identified 6 kinds of literature review methods that may support meta-analysis, when looking at how uptake of analytics affect business performance (Low-Choy, Rose & Almeida, 2021).
- At an interface with the machine learning realm, we used multiverses to explore how robust findings were to model choices, when examining how smartphone use relates to parenting (Modecki et al., 2020, 2021).
- When eliciting expert knowledge, we developed a flexible way of asking experts to assess quantities, namely risk of contamination or time to submit a thesis. We developed a bespoke Bayesian model to pool assessments by an expert to convey their mental model, then pooled these mental models across experts (Albert et al., 2012ab).
Interestingly all three publications have wide reach, being published either in a textbook or in journals with published discussion.
For further information, please contact RSFAS Seminars.
All information collected by the University is governed by the ANU Privacy Policy.