Publicly defensible
Built to withstand informed challenge, because applied work always faces it.
Applied and policy research has to be rigorous enough to withstand external scrutiny and clear enough to inform a decision. We provide the analytical depth for evidence programs where the finding will be quoted, contested, and acted upon—and where being able to show exactly how it was produced is what allows it to hold.
Research institutes and think tanks occupy a demanding position. Their work has to meet academic standards of rigor—because it will be read, and sometimes attacked, by people who know the methods—while also being timely, accessible, and useful to decision-makers who do not. It sits in public, where a methodological weakness is not just a reviewer's private note but a line of attack for anyone who dislikes the conclusion. And it often has to be produced faster, and with leaner in-house methodological resources, than a comparable academic study.
That combination creates a specific need: research-grade analytical capability, available on demand, that produces work robust enough to defend in public and documented well enough to prove. An institute cannot always justify a large permanent staff of econometricians and modellers, but it cannot afford analysis that falls apart under challenge either. When a finding shapes public debate or a policy recommendation, the cost of getting the method wrong is measured in credibility, and credibility is an institute's core asset.
We are the analytical partner for exactly this. We bring specialists in the causal, modelling, and synthesis methods that applied research relies on, apply them to your questions with full rigor, and hand back work that is reproducible from the raw data forward—so that when a finding is challenged, as public-facing work always is, you can show precisely how it was reached and why it holds.
Causal inference, impact evaluation, and modelling built to survive the scrutiny that public-facing research attracts—identifying assumptions tested, robustness established, uncertainty reported honestly.
Systematic reviews, meta-analysis, and bibliometric mapping that consolidate what is known on a question into an authoritative foundation, rather than a selective or anecdotal one.
CGE, input-output, and microsimulation for economy-wide and distributional analysis when the question is about what a policy would do, not just what has been observed.
Documented, reproducible pipelines that let you show exactly how a finding was produced when it is questioned—the difference between a result that holds under challenge and one that crumbles.
Built to withstand informed challenge, because applied work always faces it.
Academic-standard method delivered on the timelines applied research runs on.
Every finding traceable from raw data, so it can be defended when contested.
Results reported with their limits and confidence, never as false certainty.
Specialists in causal, modelling, and synthesis methods, without a permanent hire.
Sensitive or pre-release work handled under strict confidentiality.
Some engagements are single, high-stakes studies—an impact evaluation, an economic model, a systematic review—where an institute has the question and the data but needs the specialist method applied rigorously and defensibly. Here we function as the analytical engine for a flagship piece of work, doing the estimation or modelling to a standard that will hold when the finding enters public debate, and documenting it so the institute can stand behind it.
Others are ongoing relationships, where we provide a standing analytical capability that an institute draws on across its research programme—the methods bench it cannot justify staffing permanently but cannot do without. In both cases the value is the same: the institute keeps ownership of the questions, the framing, and the public voice, while we ensure the analytical foundation underneath is one that will not give way under scrutiny. When a critic challenges a finding, the institute can respond with the full, reproducible record of how it was produced—which usually ends the challenge.
Applied and policy research leans heavily on the methods where credibility is hardest to establish and easiest to attack. Causal inference is central—difference-in-differences, regression discontinuity, instrumental variables, and synthetic control—because the questions are almost always about what an intervention or a policy actually caused, and because the answers will be contested by people with an interest in a different conclusion. Economic modelling and simulation matter too, from input-output and CGE models for economy-wide impact to microsimulation for distributional effects, whenever the question is about what would happen under a policy that has not yet been tried.
Evidence synthesis is the other pillar. Systematic reviews, meta-analysis, and bibliometric mapping let an institute consolidate what is known on a question into something authoritative rather than anecdotal—increasingly the expected foundation for a serious policy contribution. We bring specialists in all of these, and we apply them with the documentation and reproducibility that applied work needs precisely because it will be scrutinized: when a finding is challenged, being able to show exactly how it was produced is what allows it to hold.
We clarify the research question, the data, and how the finding will be used and scrutinized.
We agree the right analytical approach and the deliverables, on a timeline that fits your publication or briefing.
We run the estimation, modelling, or synthesis with the diagnostics and robustness that public work demands.
We report results with their uncertainty and limits made explicit, in a form your audience can use.
You receive the reproducible pipeline—data, code, documentation—to show exactly how the finding was reached.
If a finding is contested, we help you respond with the full analytical record.
For a purely academic study, reproducibility is a matter of scientific integrity. For applied and policy research, it is also a matter of survival. When a think tank publishes a finding that cuts against a powerful interest—that a subsidy is ineffective, that a regulation has costs, that a popular policy did not work—the response is often not a counter-study but an attack on the method. The institute that can immediately show the full, documented, reproducible basis for its result can defend it; the one that cannot is left arguing from authority, which in a contested public debate is a losing position.
This is why we treat documentation and reproducibility as core to applied work rather than an optional extra. Every analysis we deliver comes with the complete record—the data as used, the code that produced every number, and the reasoning behind every choice—so that a challenged finding can be defended with evidence rather than assertion. In public-facing research, that record is not bureaucratic overhead; it is what makes the work worth publishing at all.
Institutes and think tanks publish into a contested environment, and over many projects a clear pattern emerges in which findings survive challenge and which collapse. The single biggest determinant is whether the causal claim was genuinely identified or merely asserted. A report claiming a program 'led to' an outcome on the basis of a before-and-after comparison is trivially attacked—any critic can point to the dozen other things that changed at the same time. The same claim, built on a design that credibly isolates the program's effect and reported with its identifying assumption stated and tested, is far harder to dismiss. The difference is not presentational; it is structural, and it is decided when the analysis is designed, not when the report is written.
The second determinant is how uncertainty is handled. Findings that overstate their confidence set themselves up to fail: when the single confident number turns out to be wrong, the institute's credibility takes the damage. Findings that report their uncertainty honestly—a range rather than a point, the conditions under which the result holds, the sensitivity of the conclusion to key assumptions—are both more defensible and, paradoxically, more persuasive to sophisticated audiences, who trust an analyst that acknowledges what they do not know. We build this honesty in as a matter of course, because in public-facing work it is a form of risk management as much as scientific integrity.
The third is documentation. When a finding is attacked, the institute that can produce the complete, reproducible record—the data as used, the code that generated every number, the reasoning behind every choice—can defend it with evidence, while the institute that cannot is reduced to defending it with assertion. In a public dispute, that difference usually decides the outcome. This is why we treat the reproducible record not as bureaucratic overhead but as the thing that makes a contested finding worth publishing at all, and why every analysis we deliver to an institute comes with it.
We structure engagements around the way applied research programmes actually run. The most focused is the single flagship study—an impact evaluation, an economic model, or a systematic review where the institute has the question and the data but needs the specialist method applied rigorously and defensibly. Here we function as the analytical engine for one high-stakes piece of work, delivered with the full reproducible record that lets the institute stand behind it.
The broader mode is a standing relationship: we provide an ongoing analytical capability that an institute draws on across its research programme, functioning as the methods bench it cannot justify staffing permanently. This suits institutes that produce a steady stream of evidence and need reliable, on-demand access to causal, modelling, and synthesis expertise. Between these, we also take on defined-scope components of a larger project—the evaluation module within a broader study, say—coordinated with the institute's own team. Scope, timeline, and confidentiality are always fixed in advance.
Although institutes often come to us for a specific analysis, the most productive relationships span more of the evidence lifecycle. An institute planning a major research programme benefits from involving us early—at the point of designing the studies and choosing the methods—rather than only when the analysis is due, because the credibility of the eventual findings is largely determined by decisions made at that design stage. A programme built from the outset around defensible identification and honest measurement produces evidence that holds; one that reaches for rigor only at the end is often trying to rescue conclusions the design cannot fully support.
At the other end of the lifecycle, we help institutes with the synthesis and dissemination that turns individual studies into influence. A body of an institute's own work, consolidated through a rigorous systematic review or a bibliometric map of the field it sits in, becomes a more authoritative contribution than any single study—and a stronger foundation for the policy engagement that is an institute's ultimate purpose. Working across the lifecycle in this way, rather than as a one-off analytical supplier, is where we add the most durable value: not just a defensible finding, but a research programme that is credible by design and influential by construction.
The questions clients like you most often raise before working with us.
Tell us the question and how it will be used—we'll make sure the analysis behind it is defensible.