Linking Administrative Data: An Evolutionary Schema

TL;DR

This paper discusses strategies for linking administrative data to improve statistical estimates, focusing on addressing bias and representativeness issues inherent in non-random data sources.

Contribution

It introduces a framework for handling bias and representativeness in administrative data integration, advancing methods for more accurate statistical estimation.

Findings

Developed a representativeness-based strategy for data integration

Identified bias issues in administrative data sources

Proposed components for improved estimation accuracy

Abstract

Statistics New Zealand (Stats NZ) has committed unreservedly to an administrative data first policy. Thus, all new methods used at Stats NZ are to be viewed within this context and discussing strategies for using administrative data is an integral part of every working day. As statistical methodologists, the three authors were drawn into these discussions. Like most methodologists, the authors see surveys and the publications of their results as a process where estimation is the key tool to achieve the final goal of an accurate statistical output. Randomness and sampling exists to support this goal, and early on it was clear to us that the incoming it-is-what-it-is data sources were not randomly selected. These sources were obviously biased and thus would produce biased estimates. So, we set out to design a strategy to deal with this issue. This led us to the concept of representativeness which is closely related to statistical bias but has a wider context invoking both randomness and judgement. The representativeness issue was the principal question that we set out to answer. The necessary components that we gathered for our solution are summarized in the paper. Keywords: Representativeness, Timeline Databases, Statistical Registers, Estimation