Online Program Home
My Program

Abstract Details

Activity Number: 340 - SPEED: Applications of Advanced Statistical Techniques in Complex Survey Data Analysis: Small Area Estimation, Propensity Scores, Multilevel Models, and More
Type: Contributed
Date/Time: Tuesday, July 31, 2018 : 10:30 AM to 12:20 PM
Sponsor: Survey Research Methods Section
Abstract #330414
Title: Parameter Estimate Bias Resulting from Level 3 Sample Size Decisions
Author(s): Tingqiao Chen* and Frank Lawrence and Wenjuan Ma
Companies: and Michigan State University and Michigan State University
Keywords: 3-lvl HLM; sample size; simulation; parameter bias; absolute relative bias
Abstract:

Acquiring level 3 units is more expensive than either level 1 or level 2 units (cf. Kalaian & Raudenbush, 1996). The effect of level 3 sample size decisions on level 1 parameter estimate bias is important but not known (McNeish & Stapleton, 2016). To address this shortcoming we study the impact of level-three sample size on parameter estimates when the parameters of interest were at level one. The model was a three-level linear model; the target parameter was a level-one predictor. Level-one and level-two sample sizes were fixed at 15 and 30 respectively (e.g., Kreft & de Leeuw, 1998) with variances fixed at 10 and 1. Level-three sample size varied from 5 to 30 by 5. Level-three variance ranged from 0.6 to 2.0 with ICC varying from 0.05 to 0.15 by 0.05 (e.g., Maas & Hox, 2005). Target parameter estimates were evaluated using absolute relative bias (ARB) and the parameter estimate standard error (SE). As level 3 sample size increased, both ARB and SE decreased in a non-linear fashion. The value of ARB and SE started to stabilize once the level 3 sample size reached 20.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program