Rethinking Statistical Quality in a Data-Rich World – A Comprehensive Life Cycle Framework

Thomas M. Menino Convention & Exhibition Center 

Ensuring the quality of official statistics remains a concern for National Statistics Offices. Today's statistical production increasingly relies on administrative records, blended and linked data sources, modeling, and iterative development processes that challenge traditional approaches to quality assessment. Quality can no longer be treated as a characteristic verified at the end of production; it must be embedded throughout the statistical product lifecycle. Consistent with Deming's emphasis on building quality into processes, this lecture introduces a Statistical Quality Framework that treats statistical quality as both a process and an outcome. Aligned with a Statistical Product First approach, the framework emphasizes starting with the purposes and uses statistics are intended to support and addressing quality continuously across design, development, dissemination, and evaluation—particularly when data collection is not directly controlled, and risks include representativeness, interpretability, accessibility, misuse, and context. Rather than serving as a checklist or compliance mechanism, the Statistical Quality Framework is designed to guide judgment and decision-making in the presence of uncertainty, reinforcing that quality must be built into systems and processes, not inspected in at the end.

Deming Lecture