Psychology

Special prerequisite(s):  PSC 204A

Brief course description: This course will consider modern methods for dealing with missing data in psychological studies. We will begin with an overview of common sources of missing data. We will then discuss types of missing data in the context of using common statistical methods, such as regression analysis. We will consider former treatments of missing data and discuss why these strategies yield undesirable results. Finally, we will discuss and implement modern missing data techniques (i.e., multiple imputation and full information maximum likelihood) using SAS and Mplus. Note: Students need not be familiar with SAS or Mplus for this course. We will rely on common statistical procedures, such as ANOVA and regression analysis.

List of major topics to be covered: Types of missing data, assumptions about missing data for common statistical procedures, common but problematic methods for dealing with missing data and why they often yield biased results, multiple imputation (MI), using SAS to perform MI and using SAS and Mplus to estimate ANOVA and linear regression models.

Brief description of course format: We will rely on a combination of lectures and lab exercises using computer software (SAS and Mplus). This course will be very hands-on with many examples and computer exercises.

Grading criteria: Students are expected to participate in class by engaging in class discussions (10%) and satisfactorily completing computer exercises (50%). Students will also be responsible for compiling a portfolio of course materials (lecture notes, articles, and other relevant reference materials) (40%) to be handed in at the end of the quarter.

Required readings:

Schafer, J.L., & Graham, J. W. (2002). Missing data: Our view of the state of the art. Psychological Methods, 7, 147-177.

Collins, L. M., Schafer, J. L., & Kam, C. M. (2001). A comparison of inclusive and restrictive missing-data strategies in modern missing-data procedures. Psychological Methods, 6, 330-351.

Allison, P. D. (2002). Missing Data. Thousand Oaks, CA: Sage.

Xu, S., & Blozis, S. A. (in press). Sensitivity Analysis of a Mixed Model for Incomplete Longitudinal Data. Journal of Educational and Behavioral Statistics.