The Nature of Nonresponse in a Medicaid Survey: Causes and Consequences
Patricia M. Gallagher, Floyd Jackson Fowler and Vickie L. Stringfellow
This study examines the effect of multiple phases of data collection on 1) the representativeness of the respondents and 2) survey responses to substantive items. A version of the CAHPS survey instrument was used to collect information from 1,600 adults covered by Massachusetts Medicaid about their experiences with getting health care through their health plans. In order to gain a better understanding of the significance of nonresponse, intensive follow-up with nonrespondents was done utilizing three phases of data collection: mail, telephone, and in-person protocols in successive waves of contact. Using administrative data, we were able to compare demographic and health characteristics of respondents to each phase, as well as cumulatively, with the characteristics of the total sample.
The results indicate that the addition of completed surveys from each successive phase of data collection yielded a final group of respondents that is more representative of the total population with respect to age, race, presence of a chronic condition, and annual medical costs. Responses to 40 key survey items were also analyzed by phase and cumulatively. Responses from mail and telephone respondents were significantly different for eleven of these items; responses collected from in-person interviews were significantly different from the combined data from mail and telephone respondents for three of these items.
Nonresponse, bias, Medicaid, mail, telephone, in-person interview, face to face interview, CAHPS