This paper reports the general results of research undertaken by Census Bureau staff. The views expressed are attributable to the authors and do not necessarily reflect those of the Census Bureau.

The Census Bureau conducts the Survey of Income and Program Participation (SIPP) to provide information for federal policy makers and academic analysts on topics such as poverty, income distributions, government program participation and eligibility, and health insurance coverage. The SIPP is a nationally representative, longitudinal survey. Every 10 years, we redesign the SIPP to update the sampling frame based on the most recent Census and housing units built since the Census and to make other necessary changes. Since the SIPP began in 1984, researchers and analysts have investigated and evaluated SIPP data and suggested ways to improve the SIPP sample design and weighting methodology. We have incorporated many of these ideas into the redesigned SIPP, which we will introduce in February 1996. Each SIPP panel will have a sample size of 50,000 households, be four years long, cover twelve interviews (waves), and be non-overlapping with other panels. There will be a thirteenth interview so that we can obtain a complete four calendar years worth of data and this interview will overlap with the beginning of the next panel.

We considered continuing to use overlapping panels to reduce time-in-sample bias and bias due to nonresponse which increases over the life of a panel. Time-in-sample bias occurs when response patterns change when persons are subject to repeated interviewing over the life of a panel in the SIPP. However, there was not significant evidence of such biases, so we chose a non-overlap design. (McCormick, et al, 1992.)

The SIPP has a recall period of four months, i.e. we collect data from the respondents for the previous four months. We researched the possibility of a six month recall period but determined that the longer recall period would have an adverse effect on data quality as a result of higher response variance and response bias due to the longer recall period. (Huang, et al, 1994.)

This paper gives a general description of the major areas of the redesign methodology. Section II of this paper defines the primary sampling unit (PSU) and gives an overview of the PSU stratification. We summarize the sample selection of PSUs within strata in section III and the sample selection of households within PSUs in section IV. Section V describes the weighting methodology. We conclude the paper in section VI mentioning additional redesign methodology.