|
Activity Number:
|
474
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 7, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #301268 |
|
Title:
|
An Application of Matrix Sampling and Multiple Imputation: The Decisions Survey
|
|
Author(s):
|
Mick P. Couper+ and Trivellore Raghunathan*+ and John Van Hoewyk and Sonja Ziniel
|
|
Companies:
|
The University of Michigan and The University of Michigan and The University of Michigan and The University of Michigan
|
|
Address:
|
, , , M4071 SPHII, Ann Arbor, MI, 48109,
|
|
Keywords:
|
split questionnaire ; matrix sampling ; multiple imputation ; planned missingness
|
|
Abstract:
|
We describe a split-questionnaire survey (Raghunathan and Grizzle, 1995) in a national RDD telephone survey of medical decision making among persons age 40+ in the US. Respondents who were eligible for up to 9 different medical conditions (3 cancer screenings, 3 medication, and 3 surgeries) were only asked detailed questions on up to 2 conditions. The selection of modules was inversely proportional to the expected marginal prevalence rate for each condition, and every pair of modules was administered across sampled individuals. Of the 3010 respondents, 1279 were eligible for 3 or more modules and were subject to random assignment. A total of 4528 of the 6696 possible modules were administered. We describe the selection algorithm and the subsequent multiple imputation to recover complete module-level information for all eligible modules on all 3010 respondents.
|
