Limitations of RDD telephone surveys for estimates of health care utilization: Lessons from the Community Tracking Study

John Hall, Mathematica Policy Research, P.O. Box 2393, 600 Alexander Park, Princeton, NJ 08540, 609-799-3535, jhall@mathematica-mpr.com and Mourad Touzani, Mathematca Policy Research, P.O. Box 2393, Princeton, NJ 08543.

Surveys conducted by telephone are by their nature restricted to only households with telephone service. While, a high proportion of US households have telephone service, there is still concern that the exclusion of households without telephone service may result in biased survey estimates especially where those estimates are correlated with income or other factors associated with lack of telephone service. The Community Tracking Survey (CTS) has an RDD component and a small supplemental area probability sample for households without telephones. This paper, using data from Round 3 of the CTS will compare estimates from the two components and compare combined (RDD and in person) estimates to those that would be made using only the RDD component. Analysis from the first round of CTS concluded that some estimates of health status and care utilization would be biased if interviews had not been conducted among households without telephones. Estimates of potential bias will be evaluated based on the size of the bias relative to the estimates and to their standard errors, and how the bias translates into aggregate estimates, such as persons affected and dollars spent. We will compare estimates of bias for Round 3 to those from Round 1 to see if the potential for bias increased or decreased.

Learning Objectives:

To identify issues in survey sampling design that influence the usefulness of the survey data.
To recognize limitations of telephone surveys in collecting health care utilization data from all members of the population.
To describe the segments of the population omitted from surveys that use RDD sampling and assess the potential impact of these omissions on a given survey.

Keywords: Survey, Health Care Utilization

Presenting author's disclosure statement:
I have a significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.
Relationship: Center for Health Systems Change and Mathematica Policy Research are both owned by Mathematica, Inc.

Recorded presentation

Quantitative Methods for Health Care Services and Research

The 132nd Annual Meeting (November 6-10, 2004) of APHA