Module 03 — Pattern-mining your RTI corpus

PIO replies vary wildly. Before tabulation:

1. Scan + OCR each reply (most are PDFs / scanned letters).
2. Tag with metadata: state, district, PIO name, date received, days to reply.
3. Extract numbers into a spreadsheet column.
4. Flag any partial or refused responses for follow-up.

1. Compliance speed: median + worst PIO reply time, by state/department.
2. Disclosure rate: % of sub-questions answered, by state.
3. Variance: Coefficient of variation across districts for the same metric — high CV = inconsistency / data quality issue.
4. Outliers: Top 5 + bottom 5 by metric.

1. Choropleth map (district / state level)
2. Slope chart for year-on-year change
3. Heat map for compliance scorecard
4. Bullet chart for variance vs benchmark

Avoid: pie charts, 3D bar charts (anti-pattern in data journalism).

Publish the raw RTI responses as an open dataset (CSV + scanned PDFs in a Google Drive). This:

1. Enables third-party verification
2. Encourages other journalists to dig deeper
3. Builds your NGO's credibility as a data source

License: CC BY 4.0. Tag in datasets.gov.in if relevant.

In any publication or report, cite each underlying RTI:

1. PIO designation, department, address
2. Date of RTI filing + RTI reference no.
3. Date of PIO reply + PIO reply reference no.

This citation format is what makes the story court-admissible if your evidence is later challenged.

Quiz available from your course dashboard.

• ← Course overview
• Next: Activists' RTI course - final certificate

Last reviewed: 24 April 2026.