451 – 125 Years of Representative Sampling: Important Contributions in the History of Survey Statistics
Ralph Folsom’s 1991 Social Statistics Proceedings Paper: A Seminal Contribution to Calibration Weighting for Nonresponse You Likely Never Heard Of
Jeremy Aldworth
RTI International
In their celebrated 1992 JASA paper, Deville and Särndal introduced the term “calibration weighting� to describe mild adjustments of the design weights in a survey sample that, in the absence of unit nonresponse, force weighted estimates for a vector of calibration variables to equal known population totals. Their paper showed that linear-regression weights and raking weights were examples of calibration weights. Few know, however, that in a 1991 paper in the ASA Proceedings of the Section on Social Statistics, Ralph Folsom employed what would later be called calibration equations to adjust for unit nonresponse. He showed that using raking weights leads to nearly unbiased estimated totals when the log of the probability of a unit response is a linear function of the raking variables. He also showed how to create (not-yet-called calibration weights when the probability of unit response is a logistic function of the calibration variables. Folsom and collaborators would later expand his 1991 work and develop the GEM (general exponential model) stand-alone software, which later became the basis of the WTADJUST procedure in SUDAAN.