
|
|
|
|
Conferences
|
Conferences
Warsaw 2009: Presentations and short courses
Computer-Assisted Audio Recording (CARI): Repurposing a Tool for Evaluating Comparative Instrument Design
Session: Designing and testing survey instruments for comparative research
Authors:
- Brad Edwards; Westat, United States
- Wendy Hicks; Westat, United States
- Karen Tourangeau; Westat, United States
- Lauren Harris-Kojetin; National Center for Health Statistics , United States
- Abbie Moss; National Center for Health Statistics , United States
Abstract:
The most common CARI objective (to date) has been to identify potential errors (interviewer falsification, other interviewer errors, response errors), but identification really results from CARI coupled with other tools. This paper discusses a specific application of CARI within a larger system designed to quickly identify both interviewer errors, and potential measurement error resulting from question design and implementation. Using CARI within this larger system provided the necessary tools to identify problems early in the data collection period, diminishing the potential for a negative impact on final data quality. The system has three components: audio-recording – selected questions but random subsets recorded during any one interview; behavior coding of recordings; and report generation — analysis tables generated and reviewed weekly. The system provides visibility into both interviewer and question performance; early and frequent review of information on both interviewers/questions, with flexibility in terms of amount sampled/coded and approach towards coding; and quantitative summary of results that can be verified, diminishing some of the caveats with a pure qualitative approach
Our discussion of the specific application is based on experience with a comparative establishment survey. Comparative surveys collect data from multiple groups (people living in different countries, speaking different languages, aging in different cohorts) in ways that support group comparisons. Establishment surveys are often concerned with different types of organizations. In 2007 we used CARI on the establishment-based National Home and Hospice Care Survey (NHHCS), sponsored by the National Center for Health Statistics. The NHHCS was designed to produce nationally representative data on home health care agencies and hospice care agencies. Although both agency types provide health care in the home and their services overlap, they differ in some important ways. Hospices provide palliative end-of-life care; home health agencies provide care for people with a wide range of conditions and functional limitations, for treatment and rehabilitation purposes.
|
|
|