Spaces:
Running
Running
| # User Guide | |
| ## Quick start | |
| Open the App and verify **SYNTHETIC DEMO** in the source banner. The initial example uses Chile as the treated economy, Argentina, Peru and Uruguay as comparators, a life-expectancy indicator, and a fictional analytical break in 2016. The synthetic Chile shift is engineered for demonstration and **has no historical meaning**. | |
| ### Country and time selection | |
| Choose a treated country and check at least one other country. Two or more donors are required for synthetic control. For DiD, the analysis uses the mean of *all* selected donor values in each year and excludes a year if any chosen country lacks that observation. A minimum of three observed years before and after the break is required. | |
| ### Live data | |
| Press **Load World Bank data** after choosing an outcome. The browser sends a request to the World Bank API. Data load failures leave the old data unchanged with a visible error. If you change indicator while viewing live data, press **Load World Bank data** again. The application fetches one outcome indicator at a time to keep the request small. | |
| ### Eight tabs | |
| * **Overview:** treated-versus-control means and a research design summary. | |
| * **Time series:** annually matched outcome series and the year-by-year contrast table. | |
| * **DiD:** transparent pre/post contrast, slope gap, descriptive resampling range. | |
| * **Synthetic control:** convex donor weights fitted to the pre-break years. | |
| * **Placebos:** country-reassignment and fake historical break diagnostics. | |
| * **Data audit:** observation counts, expected coverage and reporting gaps. | |
| * **Protocol:** intervention description, external source, confounder checklist and notes. | |
| * **Export:** CSV, Markdown report, and full JSON session. | |
| ### Study sessions | |
| Click **Full session .json** to download all records currently loaded, settings and protocol notes. Click **Restore a prior session** to import them back. Imported sessions are labeled **IMPORTED DATA**, even when they originated from World Bank data, since the browser cannot reauthenticate a previously downloaded JSON file. | |
| ### Interpretation | |
| The numbers are exploratory measures. The selection of countries, timing, confounding assumptions and actual policy treatment must be documented independently before any substantive impact interpretation. Read `METHODOLOGY.md` and `RESEARCH_ETHICS.md`. | |