From: Utilising identifier error variation in linkage of large administrative data sources
NHS number error distribution in simulated data | |||||
---|---|---|---|---|---|
30%, random | 0.5%, random | 30%, associated with other identifier errors | 30%, associated with ethnicity | ||
Traditional match weight | % readmitted | 7.4 | 7.4 | 6.9 | 7.4 |
Standard error | 0.002 | 0.002 | 0.002 | 0.002 | |
% bias | −15.9 | −15.9 | −21.3 | −15.7 | |
Match weight incorporating dependence | % readmitted | 7.4 | 7.4 | 6.9 | 7.4 |
Standard error | 0.002 | 0.002 | 0.002 | 0.002 | |
% bias | −16.0 | −16.0 | −21.4 | −15.8 | |
Attribute-specific match weight | % readmitted | 8.7 | 8.7 | 8.7 | 8.8 |
Standard error | 0.002 | 0.002 | 0.002 | 0.002 | |
% bias | −0.6 | −0.6 | −0.9 | −0.2 | |
Organisation-specific match weight | % readmitted | 8.8 | 8.8 | 8.8 | 8.8 |
Standard error | 0.002 | 0.002 | 0.002 | 0.002 | |
% bias | 0.2 | 0.2 | −0.1 | 0.2 |