The Role of Monte Carlo Simulation in Advancing Quantitative Methods in Psychology with Violation of Data Assumptions
Friday, March 15, 2024
2:00 p.m. - 3:00 p.m.
Online via Zoom
FREE
Conventional quantitative methods in psychology (e.g., t test, regression) are often developed based on numerous data assumptions (e.g., representative samples with no missing values, linearity, normality, homoscedasticity, etc.) that are frequently violated in psychological research practice. Hence, it is important for quantitative methodologists to examine the performance of those methods—such as bias, Type 1 error/power rate, coverage probability of confidence intervals—via Monte Carlo simulation, a computational algorithm that generates repeated random samples that violate the assumptions. A primary goal of conducting Monte Carlo simulation is to offer empirical evidence that evaluates the performance of conventional quantitative methods as well as develop new quantitative methods that are more robust (or insensitive) to violation of assumptions. In this talk, Dr. Li presents his studies that examine and address the bias arising from popular quantitative methods (e.g., group comparison, correlation, regression) with data that are subject to non-representative samples, non-linearity, and heteroscedasticity. Implication of those findings to quantitative research design, analysis and practice is also discussed.
Facilitator: Dr. Johnson Li is an Associate Professor in the Department of Psychology at the University of Manitoba, and he is serving as the Chair of the Quantitative Methods Section of the Canadian Psychological Association. His research interests include robust statistics, effect size measures, bias-corrections for study artifacts, resampling techniques, meta-analysis, structural equation modelling, and cognitive diagnostic models.
A Zoom link will be automatically emailed once registered.