InsightStack

InsightStack

Calculators that run in a browser, causal-inference code in Python and R, loaders for DHS and PLFS microdata, and templates for the documents an evaluation produces. Part of OpenStacks.

6Calculators
2Dataset loaders
28DHS surveys covered
4Languages

Calculators

Six tools · run in the browser · nothing is sent to a server

Demography

Population projection

Compound a base population forward at a constant annual growth rate.

Open →

Education

Education needs

Project school-age population, then compare schools required against schools that exist.

Open →

Finance

Financial access gap

Count banking access points against the number a district's population implies.

Open →

Health

District health indicator

Adjust a state NFHS figure towards a district with three coarse multipliers.

Open →

Disaggregation

Block-level estimator

Split a district count down to one block by its share of district population.

Open →

Environment

Environmental fragility

Score four indicators against thresholds and weight them into one composite.

Open →

Each calculator states its assumptions on the page. Where a published district figure exists, use that instead.

Code and tooling

Technique, templates and worked examples

EconometricsPython, R Difference-in-differences, matching, instrumental variables, regression discontinuity and sensitivity analysis, with sample data. Stata snippets44 do-files Data management, merges, labels, graphs, regressions, survey settings. SPSS scripts20 files Survey analysis for teams whose institution runs on SPSS. Network effectsPython Centrality, peer association with leave-one-out means, and threshold diffusion for self-help group data. ReplicationPython, R One command from raw data to results, a recorded answer, and an R cross-check. Data validationPy, R, Stata, SPSS Checks from a data dictionary, reported by household id: duplicates, required fields, ranges, allowed values, types, column set. Label variablesPy, R, Stata, SPSS Variable and value labels from a dictionary, written into Stata and SPSS files. Survey to codebookPython An XLSForm to a Markdown codebook and a label dictionary. Reports form defects such as a missing choice list. System dynamicsVensim Eight system models across health, agriculture, climate, migration and education. Tool notes9 tools Excalidraw, Kumu, Observable, RawGraphs, Flourish, Power BI, Miro, Excel and LaTeX: when to use each, when not to, and one example file apiece. Qualitative codingTaguette One focus group coded start to finish in Taguette, with the method. Recorded fieldworkDescript Recorded interviews: consent, transcription, clips and quotes. Annotated research5 briefs Five illustrative research briefs annotated with one eight-tag scheme, weakest design to strongest. Learning libraryReference A reading shelf: programming, data science, research methods, AI tools. Learning layersPractice Learning in the MEL cycle as scheduled decisions, at three tempos. Writing guides5 guides Theory of change, results chain, evaluation report, policy brief, writing about uncertainty. Evaluation documentsTemplates Logframe, indicator reference sheet and MEL framework templates, with a completed example. Knowledge managementConventions A folder structure, a README template, file naming and tagging conventions.

Links open on GitHub.

Status

Stable

The code works and is not under active development. Bug reports are welcome; new features are unlikely, and replies take weeks. See the maintenance policy.

DOI 10.5281/zenodo.15245182. Companion repositories: FieldStack for R and EquityStack for Python.