Skip to main content

RE: weighting

HUD.GOV HUDUser.gov
eList
AHS
Richard (and others who may be interested)-

Generally, if you are comparing across years (a cross-section), you should be using the WEIGHT variable. That weight variable represents different households and is adjusted (calibrated) each year based on other surveys. That is the best one to look at if you are creating estimates. However, these weights change over time, so you’ll get different values depending what year you use.

However, if you really are concerned about the same households over time (using it longitudinally), then you might want to use the pure weights (PWT), which is the inverse of the probability of selection from the original sample being drawn. However, these won’t match control totals printed in the reports.

In the Components of Inventory Change (CINCH) reports (available from HUD USER), there is a description of the weighting procedure used there. I also can make available the presentation from the last Housing Statistics Users Group (HSUG) meeting where I discussed some of the weighting issues in creating CINCH.

That’s the very short answer, the much longer answer involves explaining where there are a lot of differences and pros and cons of different things. Dav Vandenbroucke and Ron Sepanik from HUD have been involved in this for years, as have Dennis Schwanz, Paul Harple, and Barbara Williams from Census. All of them may also want to weigh in. It is not an easy topic.

Greg


Gregory Watson
Vice President
ICF Consulting
703-934-3694 (v)
703-934-3156 (f)
gwatson@icfconsulting.com

ICF Consulting has, and is, supporting HUD’s Office of Policy Development and Research as a contractor. All opinions are that of ICF Consulting only.
NOTICE:

This message is for the designated recipient only and may contain privileged or confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of this e-mail by you is prohibited.