Do We Learn from Past Experience When Constructing Complex Data

The Survey of Income and Program Participation (SIPP) is a nationally representative longitudinal survey of households in the United States, designed to alleviate a number of problems with currently available socioeconomic data. The survey content and collection methods provide the ability to measure intrayear fluctuations in transfer program eligibility and participation, net worth, income, expenses, employment, and household composition.l SIPP grew out of the Income Survey Development Program initiated by the Department of Health and Human Services in the mid-1970s in response to the need for an improved measure of the economic situation of households in this country. The Income Survey Development Program sponsored extensive research into ways in which the measurement, collection, and processing of income, transfer program, and wealth data could be improved. In the late 1970s several site tests and two nationally representative research panels were fielded in order to test alternative collection and processing methods. The last of these tests, known as the 1979 Income Survey Development Program Research Test Panel (ISDP), was sufficiently large to provide reliable national estimates of many household characteristics.2 However, the survey was so complex and so new that there were major hurdles to overcome before the data could be used for analytical purposes. The objective of this paper is to compare the public use microdata products from SIPP to those available from the ISDP. Areas where improvements have been made will be discussed as will areas where more work should be done to facilitate the use of SIPP for the analysis of public policy, particularly that policy which affects the low income population. The paper first provides a brief history of the ISDP and lists a series of difficulties experienced with the ISDP from the perspective of a data coordinator whose job was to generate analysis files for specific research tasks and to assist researchers in understanding the data and its limitations. Following that is a section describing the areas in which improvements have been made to SIPP to minimize these difficulties. The paper concludes with the author's "wish list" for further enhancements to the SIPP data products. I might note before proceding that the answer to the question posed in the title of the paper is yes, much has been learned from the ISDP experience.