Teaching Missing Data Methods through Monte Carlo Simulation

• In Event: Simulation as Method, Pedagogy, and Epistemology: A Framework for Monte Carlo Simulation Methods

Abstract

This presentation explores how Monte Carlo simulation can serve as both a methodological tool and a pedagogical strategy for teaching applied topics in quantitative research. This example focuses specifically on the complex area of missing data analysis, the speaker presents an instructional module designed within a semester-long simulation course that engages students in designing and interpreting simulations to understand missingness mechanisms and compare estimation strategies.
The module introduces students to key missing data concepts, Missing Completely at Random (MCAR), Missing at Random (MAR), and Missing Not at Random (MNAR), and explores their implications through simulation studies that contrast techniques such as listwise deletion, mean imputation, and Full Information Maximum Likelihood (FIML). Using the 6 C’s framework (Conceptualize, Comprehend, Conduct, Calculate, Chart, Communicate), students begin by identifying substantive research questions affected by missing data, then simulate datasets under varied missingness conditions to examine bias, efficiency, and convergence of parameter estimates. Emphasis is placed on linking theoretical knowledge with simulation evidence to develop robust intuitions about model behavior under different data conditions.
The instructional design scaffolds learning across multiple stages. Early class sessions focus on building comprehension of statistical assumptions and preparing simulation syntax. Mid-module activities involve structured in-class coding walkthroughs and guided questions to analyze simulation output. Final deliverables require students to synthesize findings into reproducible reports, using Quarto or RMarkdown, that communicate implications to both technical and applied audiences. In addition to technical fluency, the module fosters reflective thinking about data quality, analytic transparency, and the interpretive consequences of methodological decisions.
This presentation provides code samples, assessment prompts, and instructional scaffolds used in the module, along with reflections on student learning outcomes. It also discusses pedagogical strategies for balancing flexibility and support, particularly when teaching students with diverse backgrounds in statistics and programming. By integrating simulation directly into the teaching of a substantive methodological topic, this paper demonstrates how simulation can deepen understanding of statistical principles while promoting reproducibility and epistemic humility.

Author

• Kirsten Reyna, University of Connecticut