A.C.E. Revision II
December 31, 2002
Evaluation of At-Risk Codes
Tamara S. Adams
Elizabeth A. Krejsa
Demographic Statistical Methods Division
Planning, Research, and Evaluation Division
What is the primary goal of the At-Risk Codes evaluation?
Since the evaluations of the Accuracy and Coverage Evaluation Survey (A.C.E.) found errors in
the assignment of enumeration and residence status of both the E-sample and the P-sample, the
revision of the A.C.E. included recoding a subsample of the A.C.E. sample and using the results
in a double sampling ratio adjustment. The recoding operation for the A.C.E. Revision II
assigned some of the E-sample enumeration status codes and some of the P-sample residence
status codes by a computer algorithm and the rest by analysts at the National Processing Center
(NPC). The primary goal of the At-Risk Codes evaluation is to estimate the potential error in the
A.C.E. Revision II dual system estimates (DSEs) due to the automated assignment of
enumeration and residence status for some of the cases. Since the only portions of the DSE that
involve the revision coding are the double-sampling ratios, we will concentrate our analysis on
What did the At-Risk Codes evaluation find in the E-sample?
Twelve of the 31 E-sample double-sampling ratios have significant differences between the
A.C.E. Revision II and those with the at-risk adjustment. The differences in the double-
sampling ratios range from –0.0025 (se=0.0016) to 0.0009 (se=0.00004). The largest increase in
a poststratum due to the at-risk adjustment is an additional 8,743 (0.035%) correct enumerations;
the largest decrease in a poststratum due to the at-risk adjustment is a decrease of 10,798
(0.246%) correct enumerations.
What did the At-Risk Codes evaluation find in the P-sample?
In the P-sample, none of the double-sampling ratios with the at-risk adjustments were