- Browse
Search
Program Calendar
Browse By Day
Browse By Time
Browse By Person
Browse By Room
Browse By Unit
Browse By Session Type
Browse By Descriptor
- Information Menu
Search Tips
Annual Meeting Housing and Travel
- Navigation and Settings Menu
Personal Schedule
Sign In
- Social Media Menu
Facebook
X (Twitter)
The Effects of Fully Conditional Specification for Treating Missing Data in Multilevel Regression Modeling With Unequal Group Size
Mon, May 1, 8:15 to 9:45am, Henry B. Gonzalez Convention Center, Floor: Ballroom Level, Hemisfair Ballroom 2Abstract
Fully conditional specification (FCS) is an increasingly used multiple imputation approach to treat missing data with nested structure. However, practices associated with the employing of FCS on missing data in multilevel regression modeling (MRM) are problematic due to the lack of empirical research. Also, little is known about how the complexity of Level-2 missingness and unbalanced group size affect FCS’s performance. A Monte Carlo simulation was employed here to investigate the effects of FCS to treat level-2 missing data in MRM with unbalanced groups. The results showed that FCS performed well in most conditions, but is currently not suitable for recovering parameters and capturing random effects in low ICC, sample group size combined with high missing rate conditions.