Respondent selection by mail: obtaining probability samples of health plan enrollees

This study was done to assess the feasibility of respondent selection by mail to obtain random samples of both child and adult enrollees of health plans when only subscriber contact information is available.

The subjects were enrollees of health plans covered under the policies of employees of the State of Washington. Subscribers were eligible for inclusion in the study samples if they had been enrolled for at least 6 months and, depending on the test, had at least one child and/or a spouse enrolled under their policy. Subjects were randomly assigned to six groups: three approaches for sampling children, two approaches for sampling adults, and one test of parents' willingness to return a questionnaire about themselves. Child selection protocols involved asking respondents to complete either a child-only or a dual questionnaire, asking them to follow a decision rule to choose a sample person, and collecting data in two phases, asking respondents to return material twice.

Results indicated that asking subscribers to select an adult respondent by mail was not a success. At least given the procedures we used, adults did not demonstrate good compliance with the respondent selection process offered them. In contrast, parents proved willing to follow a more complicated child selection rule and to do it nearly perfectly.

Parents will follow a decision rule to select an eligible child, but requiring this additional respondent selection step may be associated with a slightly decreased response rate. Asking parents to return materials twice is feasible, but it is too cumbersome a procedure to be practical. It is possible to collect data about both an adult and a child using a dual instrument; however, the increase in information is tempered by a decrease in response rates.