Use of Dependent Interviewing Procedures
to Improve Data Quality in the Measurement of Change
Jeffrey Moore, Nancy Bates, Joanne Pascale, Julia Klein Griffiths, Aniekan Okon
KEY WORDS: longitudinal/panel surveys, measurement error, questionnaire design, seam bias, transition data
ABSTRACT
Panel surveys often – and perhaps always – suffer to some extent from seam bias, the tendency for estimates of change measured across the “seam” between two successive survey administrations to far exceed change estimates measured within a single interview. Much research has documented the existence of seam bias; attempts to reduce it, however, have met with only limited practical success. The U.S. Census Bureau recently completed a multi-year research program to improve the questionnaire for the Survey of Income and Program Participation (SIPP). One important goal of that program was a significant reduction in seam bias; the questionnaire revision which was intended to accomplish this was a much more extensive and focused use of dependent interviewing (DI) procedures. This paper describes the new DI procedures, and examines their impact on estimates of month-to-month change across the first several waves of the new panel for a number of characteristics (e.g., participation in government transfer programs, school enrollment, health insurance coverage, employment, and earnings amounts) through a comparison with similar estimates derived from 2001 SIPP panel data. We find clear evidence of improved quality in the measurement of change with the new procedure – estimates of month-to-month change from the initial interview waves of the 2004 panel are much less afflicted with seam bias than their 2001 counterparts. Even with the improvements, however, we find that much seam bias still remains in the new estimates – but the findings also highlight some potentially fruitful avenues for achieving additional gains in the future.
CITATION:
