---
title: Evidence Reviews
group: Research
order: 5
summary: Import citation records, freeze a local protocol, reconcile screening decisions and export traceable review inputs alongside narrative extractions.
status: live
statusNote: A private local study register connects the bibliography to protocol versions, recorded screening and extraction inputs. Reviewer labels are self-recorded. Actual searches, independent source checks and systematic-review publication require people and evidence.
conditions: [measurement-gap]
firstMilestone: Reviews of 25 widely funded interventions, each graded and checked against its sources by a second contributor.
help:
  - Write a review of an intervention you know well, following the Intervention Database format, and open a pull request.
  - Check the sources of an existing review and note in the pull request what you verified.
  - Open an issue labeled program naming an intervention that needs a review.
related: [/record/interventions/, /record/evidence-library/, /programs/replication-program/, /method/]
sources:
  - title: About us
    publisher: Cochrane
    url: https://www.cochrane.org/about-us
  - title: The Campbell Collaboration
    publisher: The Campbell Collaboration
    url: https://www.campbellcollaboration.org/
---

## Use the live research desk

Use the study register to freeze a local question and protocol, then preview references from the [Evidence Library](/record/evidence-library/) or a mapped CSV. Inspect proposed duplicates before combining reports. Preserve the original citation rows, recorded search provenance, screening reasons and protocol/source revisions.

Record title/abstract decisions in two labeled lanes, resolve a disagreement with a reason, record whether a full report was obtained, then screen that report. Both lanes can be entered by the same person. The software counts included reports and unresolved work exactly; it establishes no independent review or unique-study count. Optional extraction inputs preserve an estimate, unit, interval, source reference and unknown fields without pooling effects.

The existing [narrative extraction desk](#workspace-evidence-reviews) remains available for intervention-specific question, bias and synthesis notes. Its earlier saved records and key are unchanged. There is no automatic migration or approval between the two workspaces.

Your workspace stays in this browser. Save records, edit them or fork frozen versions, export JSON, a research memo or CSV, and import a backup with an explicit replacement step. Deleting or clearing offers an undo while the page remains open. No notes are submitted to Charitology. The institutional goals below remain the program’s longer-term charter.

## What it is

A systematic review gathers the good studies on a question, judges how strong each one is and reports what they show together. Evidence Reviews applies that discipline to the interventions charity pays for, and publishes each review in the [Intervention Database](/record/interventions/).

## How it works

Every review follows one format. It says what the intervention does and how it is supposed to work. It grades the evidence as strong, moderate, mixed, weak or unknown. And it records where the work sits on the [depth scale](/method/#depth), how soon its effects arrive, how long they last and who designed it: the community, the community with outside experts, or outsiders alone. Where cost data exists, the review gives a cost range with its date and source.

The program builds on reviews that already exist instead of redoing them. Cochrane, a global network of health researchers, professionals, patients and carers, publishes systematic reviews in health. The Campbell Collaboration produces evidence syntheses for economic and social policy. Where one of their reviews covers an intervention, Evidence Reviews summarizes it in plain language and links to it. Where none exists, contributors write a lighter review and say so.

## Why it matters for the system

The [Measurement Gap](/diagnosis/measurement-gap/) concerns the difference between counting activity and estimating its effects. A small charity may benefit from evaluation partners and shared research. If a good review shows that an intervention works in settings like its own, the charity can borrow that knowledge instead of building it from scratch. Shared reviews turn measurement from a cost every organization pays separately into a common good the whole field can use.

Reviews also make the gaps visible. An intervention graded unknown is a finding, and a target for the [Replication Program](/programs/replication-program/) or for new research.

## Risks and limits

Evidence from one place doesn’t always carry over to another. Each review should say where its studies were done and who took part, so readers can judge how far the results travel. Reviews also go stale as new studies appear, so each one should say when it was last checked.

## What success looks like

Within five years, the most widely funded interventions each have a current, graded review, and donors and charities cite those reviews when they decide what to fund.