Policy Organizations

Policy evidence stands or falls on two things: whether a causal claim is actually identified, and whether its uncertainty is honestly represented. On the first, we work in the full toolkit of modern impact evaluation—randomized controlled trials where they are feasible, and quasi-experimental designs (difference-in-differences, regression discontinuity, instrumental variables, matching, synthetic control) where they are not—always with the identifying assumptions tested rather than asserted. A policy conclusion built on a design whose assumptions were never checked is not evidence; it is an opinion with regression output attached, and it will not survive informed challenge.

On the second, we use economic and simulation modelling to project effects that cannot be observed directly—the economy-wide consequences of a tax change, the distributional impact of a benefit reform, the range of outcomes under an uncertain future—and we report them with Monte Carlo and sensitivity analysis so the uncertainty is explicit rather than hidden behind a single point estimate. Cost-benefit, cost-effectiveness, and budget-impact analysis translate those results into the terms decisions are actually made in. The common thread is that everything is built to be defensible in public, because policy evidence always is.

The value of policy evidence depends entirely on its independence. Evidence produced to justify a decision already made is not evidence; it is advocacy with statistics attached, and sophisticated audiences recognize it as such. We therefore work as an independent analytical partner, and we report what the analysis actually shows—including, and especially, when that is not the conclusion that was hoped for. An evaluation that finds a favoured program did not work is doing exactly its job, and we deliver that finding as clearly as any other.

This independence is not a constraint on the value we provide; it is the source of it. Policy evidence that can be trusted precisely because it was produced honestly—that reports null results, that states its uncertainty, that survives scrutiny—is worth immeasurably more to a serious organization than evidence engineered to please. We hold that line because it is the only basis on which policy analysis is worth commissioning at all, and because a reputation for honest analysis is, in the end, the thing that makes an organization's evidence carry weight.

Policy engagements are shaped by the decision they inform and the timeline it runs on. The most common is a commissioned evaluation or modelling study: an agency needs credible evidence on whether a program worked, or a defensible projection of what a proposed policy would do, and we deliver that analysis to academic standard but on a policy timeline. Because these findings enter public debate, they come with the full documented, reproducible basis needed to defend them under challenge.

We also support organizations on an ongoing basis, as an independent analytical partner they can call on across a portfolio of decisions rather than for a single study—useful for bodies that regularly need rigorous impact evidence but do not maintain a large in-house methods team. And we take on independent-review roles, providing an external check on evaluations or models produced elsewhere. Across all of these, our independence is the point: we agree scope and terms in advance, and we report what the evidence shows regardless of which conclusion was hoped for.

For a policy organization, the value of rigorous, independent analysis compounds well beyond any single decision. An agency that consistently produces evidence which survives scrutiny—that reports null results honestly, states its uncertainty, and can defend its methods when challenged—builds a reputation for credibility that makes all of its subsequent analysis carry more weight. Conversely, a single high-profile finding that collapses under challenge damages trust in everything the organization produces afterwards. Because we understand that the stakes extend beyond the immediate question to the organization's standing, we hold to the same rigor and honesty on every engagement, and we help build the documented, reproducible record that lets an organization defend its evidence and, over time, earn the trust that makes it persuasive.

This long view also shapes how we handle the relationship between analysis and advocacy. Policy organizations often have positions, and there is nothing improper about that—but the credibility of their evidence depends on the analysis being conducted independently of the desired conclusion. We help maintain that separation, producing evidence that can be trusted precisely because it was not engineered to support a predetermined answer. An organization that can point to genuinely independent analysis behind its positions is far harder to dismiss than one whose evidence is suspected of following its advocacy, and preserving that distinction is one of the more valuable things an external analytical partner provides.