Discussions

Discussions

Ask a Question
Back to all

Rental data time series variation

3 months ago

Hi Dewey team,

We're analyzing rental listing data for four pilot states (VA, VT, WA, WI) and noticed an interesting pattern in the time series of monthly observations:

  • Pre-2018: Observation counts are notably higher across all four states
  • Late 2018: Sharp drop in observation counts (particularly visible in VA and WA)
  • Post-2018: Counts stabilize at lower levels

Our questions:

  1. Is this pattern expected based on changes in Dewey's data collection methodology or coverage?
  2. Did something specific change in late 2018 regarding data sources, scraping frequency, or geographic coverage?
  3. Should we be concerned about comparability of pre-2018 vs post-2018 data for panel analysis?

Understanding this will help us decide whether to:

  • Use the full 2014-2025 period with appropriate controls
  • Focus analysis on the more stable post-2018 period
  • Apply different treatments to the two periods

Thank you for any insight you can provide.