245 – Working with Missing Data: Nonresponse, Imputation, and Suppressed Data
An Iterative Cutoff Sampling Method Applied to EIA’s Annual Survey of Domestic Oil and Gas Reserves
Jason Worrall
U.S. Energy Information Administration
Samson Adeshiyan
U.S. Energy Information Administration
The Energy Information Administration's survey Form EIA-23, "Annual Oil and Gas Reserves Report" is an annual survey of oil and gas well operators in the United States, collecting information on oil and gas reserves. Historically the survey was administered to a certainty stratum, plus a Probability Proportional to Size (PPS) stratum. The sample has been changed to a cutoff sample, due to the fact that item response rates among the smaller operators were poor. The advantage of cutoff sampling is in using the smallest necessary sample to get useful data in a highly skewed population. This population features a high level of heteroskedasticity and heterogeneity. When estimation groups and publication groups are non-overlapping, determining appropriate cutoffs can be challenging. An iterative sampling approach was used to determine cutoff levels by region to achieve target Relative Standard Errors. Additionally, an evaluation of model failure is considered in the absence of any historical reserves data for smaller respondents.