time as states experiment with different approaches. It will be important to develop ways for national household surveys to keep up to date with changes in programs.
Modifying the sampling schemes of existing surveys is important to meet data needs for program analysis and monitoring. The devolution of responsibility for social welfare programs from the federal to state governments implies a greater need for state-level estimates. Currently, none of the national surveys provides complete and reliable estimates at the state level. Similarly, since program eligibility criteria may differ for population groups (e.g., immigrants), the existing samples may need modification to provide reliable estimates for particular groups.
Even with improved survey data, researchers face methodological challenges in assessing the effects on program participation and other behaviors of major changes in programs. Altered survey content that affects the comparability of measurements is one problem that may be hard to avoid; there is also the problem of estimating how respondents would have behaved if the programs had not changed.
The workshop was organized to address two general issues—how to keep national household surveys relevant for program monitoring and analysis and how federal program and statistical agencies can work together more effectively to this end—but workshop participants also acknowledged that other agencies and other data sources are important in an era of program devolution. Thus, partnerships of federal and state agencies will be vitally important, as will communication channels among public agencies, academic researchers, and private organizations, to ensure adequate data collection and analysis of the nation's health and social welfare programs in an era of change.
The workshop participants recognized that providing adequate national data for health and social welfare programs in the new environment will involve tradeoffs. For example, data needs may suggest lengthening survey questionnaires, but longer questionnaires may impair public cooperation. Also, limited resources may force difficult choices between expenditures for validation of new questions and expenditures for increased sample sizes, or between expenditures for new questions on programs and expenditures for existing questions on other topics. These and other tradeoffs will require careful consideration by statistical and program agencies. The need to make choices and to develop innovative ways to maximize the use of scarce resources underscores the important role of the newly established interagency coordination mechanism for improving national surveys for health and social welfare programs.