The evidence paradox – or when is a series not a series?

In their recent briefing paper on breaks and discontinuities in official data series in the UK, Ruth Dixon and Christopher Hood highlighted the tension between the demand for quantitative evidence to drive performance improvement and the tendency to systematically destroy the very evidence by which performance can be evaluated. This paper was discussed, and further examples of data breaks across the public sector were explored, at a seminar at LSE in April 2015 attended by senior civil servants and academics. An ensuing discussion embodied the same tensions, with some participants emphasising the need for indicator continuity, and others stressing that indicators must change as methodologies, purposes, and audiences evolve. Can this tension be resolved? In this article, Ruth Dixon, Tony Travers and Christopher Hood suggest that recommendations arising from the seminar might point to a way to reconcile these demands