13 · The data landscape

Tuesday, Oct 6, 2026

Materials for this session are not published yet. They appear here before class.

Objectives

By the end of this session you can:

• Map an economic question to candidate data families.
• Apply the granularity / frequency / signal checklist to a research idea.
• Classify a data source’s access class.
• Name each acquisition channel and when it applies.

What we cover

• Part II opens with the problems → data map: growth measured from nightlights, policy from text, history from scans, monetary policy from vocal tone, environment from satellites, prices from scraped listings.
• The six data families, and the sources triad: open, licensed, self-collected — plus replication packages.
• Channels and licensing; what we skip (video, networks) and why.
• The formats data actually arrives in, one line each: Excel files (readxl), Stata files (haven), R’s own .rds, semicolon CSVs (read.csv2), and fread() when the file is big.
• The requirements checklist: at what granularity, at what frequency, and in which fields does your phenomenon leave a signal?

Verification habit. For any question — where would the signal live?