This paper investigates “seam bias” – a tendency for reported change to occur much more frequently in the “seam” between two successive survey waves than between months within the reference period of a single survey wave – in the Income Survey Development Program (ISDP), the precursor to the Survey of Income and Program Participation (SIPP). It finds clear evidence of the seam effect across virtually all examined characteristics. The paper discusses several possible mechanisms for seam bias, including real underlying trends, edit and imputation procedures, mismatched persons across survey waves, and response error. It concludes that, of these possible factors, only the first is unlikely.
Moore, Jeffrey C. and Daniel Kasprzyk. (2010).* Month-to-Month Recipiency Turnover in the ISDP. Statistical Research Division Research Report Series (Survey Methodology #2010-03). U.S. Census Bureau. Available online at <http://www.census.gov/srd/papers/pdf/rsm2010-03.pdf>.