|
Activity Number:
|
468
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #308150 |
|
Title:
|
Examining Benefits and Drawbacks of Multiple Imputation for Disease Rate and Demographic Data
|
|
Author(s):
|
Irene Helenowski*+ and Hakan Demirtas and Chiang-Ching Huang
|
|
Companies:
|
Northwestern University and University of Illinois at Chicago and Northwestern University
|
|
Address:
|
680 N Lake Shore Drive, Chicago, IL, 60646,
|
|
Keywords:
|
multiple imputation ; spatial statistics ; parameter estimation
|
|
Abstract:
|
Missing data could be prevalent in spatial problems, as in geographic studies and examining spatial reproducibility of biomarkers in a particular organ. Multiple imputation methods based on the sample distribution of observed data or bootstrapping predicted values from a regression model have been used to circumvent this problem. In this work, plausibility of the imputation approach is examined via measures of spatial discrepancies calculated through multiply imputed datasets. We examine the reliability of the multiple imputation methods described above to study the impact of demographic factors on cancer and cardiovascular disease rate in different geographic areas by examining estimated regression parameters with commonly accepted accuracy and precision measures that involve coverage rate and average width for different numbers of imputed datasets.
|
