Sponsoring Section/Society: ASA Program Chair Session
Session Slot: 10:30-12:20 Thursday
Estimated Audience Size: xx-xx
AudioVisual Request: TBA
Session Title: Memorial Session: James T. Massey
Theme Session: No
Applied Session: Yes
Session Organizer: Zell, Elizabeth R.
Session Organizer: Ezzati-Rice, Trena M. CDC
Address: 1600 Clifton Road, MS-E52
Email: firstname.lastname@example.org and email@example.com
Session Timing: 110 minutes total (Sorry about format):
Opening Remarks by Chair - 5 or 0 minutes First Speaker - 30 minutes (or 25) Second Speaker - 30 minutes Third Speaker - 30 minutes Discussant - 10 minutes (or none) Floor Discusion - 10 minutes (or 5 or 15)
Session Chair: Zell, Elizabeth R. CDC
Address: 1600 Clifton Road, MS-E62
1. Using the National Health Interview Survey as a Sampling Frame for Other Health-Related Surveys
Ezzati-Rice, Trena M., NCHS
Address: 6525 Belcrest Road; Hyattsville, Maryland 20782
Phone: 301-436-7022 x133
Cohen, Steven, AHCPR
Abstract: A number of important health and health care surveys are carried out by the Department of Health and Human Services (DHHS). A major priority of the DHHS Data Council over the past several years has been the movement toward integrated, coordinated data systems that can meet multiple health data needs in an efficient and linked framework. The linkage of data systems through the use of common core questionnaires and sampling frames is designed to enhance analytic capabilities, maximize survey efficiencies, and minimize survey costs. The National Health Interview Survey (NHIS), a continuous survey of the U.S. population conducted by the National Center for Health Statistics, is an important component of the nation's health statistics data collection system. DHHS conducts a number of other health and health-related surveys. This paper will highlight the key features of linking the design of three major population-based surveys to that of the NHIS. The use of NHIS samples from prior years as screening samples to find domains of special interest is one advantage of linkage. Cycle IV of the National Survey of Family Growth (NSFG) was the first periodic household survey to use the NHIS as a sampling frame. The NSFG, a survey of women ages 15 to 44 with oversampling of Hispanic and black women, is designed to produce estimates of factors related to pregnancy and the health of women and infants. A redesigned National Health and Nutrition Examination Survey (NHANES IV) will move to a continuous survey and will use NHIS as a sampling frame. Finally, an important longitudinal panel survey on insurance, expenditures, health care utilization, and sources of payment, The Medical Expenditure Panel Survey (MEPS), conducted by the Agency for Health Care Policy Research, also uses the NHIS as its sampling frame. The advantages, as well as some of the issues and limitations associated with the design of these linked surveys, such as respondent burden, tracking, response rates and use of the NHIS sample for multiple surveys will also be discussed. Finally, we will outline potential methodological evaluations for the next redesign of the NHIS as it relates to linked surveys in the future.
2. Telephone Surveys: Past, Present, and Future
Cox, Brenda G., Mathematica Policy Research, Inc.
Address: 600 Maryland Ave SW, Suite 550; Washington, D.C. 20024
Mitchell, Susan, Mathematica Policy Research, Inc.
Abstract: The past 20 years has seen the telephone become the predominate data collection mode for publically and privately sponsored surveys. Paying special attention to the contributions made by James Massey, this paper describes critical milestones in the recognition of telephone surveys as a valid data collection approach and the challenges becoming apparent for the future. Massey and Thornberry were one of the first to document the dramatic improvement in telephone coverage for the U. S. population. This improved telephone coverage together with its low cost led survey sponsors to substitute the telephone for more expensive personal interviews or less reliable mail questionnaires. Simultaneously, researchers initiated investigations designed to identify ways to improve the quality of telephone survey data. Methodological research was conducted to determine methods to improve response rates, to describe mode differences between telephone and face-to-face surveys, and to delineate response category ordering effects. This research together with the widespread adoption of computer-assisted telephone interviewing (CATI) in the late 1970's and early 1980's led to continuing improvement in the quality of telephone survey data. The last decade has seen increasing emphasis on reducing the cost of telephone interviews while maintaining data quality. Methodologists now focus on issues such as optimal calling times and patterns, the effect of interviewer and respondent incentives, and automated approaches for call scheduling and dialing. Samplers have shifted from the classic Mitofsky-Waksberg approach for random digit dialed surveys to list-assisted sampling, reducing survey costs while making data collection easier to manage. As our final topic, we describe what we see as future directions for telephone surveys. Samplers need to explore the use of telephone surveys as an efficient screening tool for rare populations and, in combination with face-to-face interviews, as an affordable method when state-level estimation of low income populations is needed. Methodologists need to pursue parallel investigations to uncover methods to overcome interviewing barriers posed by innovations such as answering machines and automatic caller identification as well as how to take advantage of new technologies such as touchpad interviewing, predictive dialing, and decentralized interviewing.
3. An Evaluation of the Bias in Telephone and Face-to-Face Interviewing for the National Health Interview Survey
Biemer, Paul P., Research Triangle Institute
Address: 3040 Cornwallis Road; Post Office Box 12194; Research Triangle Park, North Carolina 27709
Abstract: This study examines the quality of RDD data relative to data collected by face to face interviewing in the National Health Interview Survey (NHIS). The motivation for the study is to determine the feasibility of combining RDD and NHIS data in order to reduce the variance of state-level estimates. Currently, since NHIS is strictly a national survey, there are only a few states where the annual NHIS sample sizes are adequate to produce state-level estimates of the required precision. If this supplementation of the NHIS sample with state-based RDD samples proves to be feasible, dual frame survey estimates of NHIS characteristics could be an option for states who need state-level estimates of greater precision than can be produced by the NHIS.
The primary determinant of the feasibility of this dual frame, dual mode approach for the NHIS is the relative quality of the RDD data. The present study was designed to evaluate the relative magnitudes of the nonsampling error components associated with the RDD survey and the NHIS. Two states were selected for the study: Texas and California. In each of these states, about 2000 interviews were conducted by CATI and an equal number of interviews were conducted for the regular NHIS. Further, about 500 RDD households and 1000 NHIS households were reinterviewed in each state within three weeks of the original interview using replicate reinterview procedures. Finally, a follow-up of NHIS nonrespondents was conducted netting a total of 214 respondents (40% nonresponse conversion rate).
In this paper, we describe the results of a latent class analysis of the combined NHIS and RDD/NHIS data. Estimates of measurement and nonresponse bias are presented and compared for both modes. Using these estimates, we compare the biases of RDD CATI and face to face administration for the NHIS. We conclude that, when the effects of nonresponse bias and measurement bias are combined, the data quality for the telephone mode does not differ appreciably from that of face to face interviewing for key NHIS data items. Our general results have implications for most surveys conducted by centralized, computer assisted telephone interviewing.
Discussant: White, Andrew A. National Research Council
Address: 2101 Constitution Ave. N.W., Rm. HA192 Washington, D.C. 20418
List of speakers who are nonmembers: None