Posted Date: April 24, 2019
HUD User elist
AHS banner

Attention AHS users: Over the next few days, HUD and Census will make three announcements about major updates to AHS products for 2011, 2013, 2015, and 2017. We encourage you to read each of these three messages carefully as they each contain valuable information!

This first message announces some important changes to the 2011 PUF.

A brief history of the 2011 and 2013 AHS PUFs

Before 2011, we always published the AHS national sample (AHS-N) data and the metropolitan area oversamples (AHS-MS) data as separate PUFs. The only exception to this practice was the AHS-MS data from the "Big 5" areas (Chicago, Detroit, New York City, Northern New Jersey, and Philadelphia) and data from a sixth "Big" area (Los Angeles). For 2003 and 2009, we integrated the Big 5 metropolitan area oversamples into the existing AHS-N data to produce the AHS-N PUF. For 2007, we integrated the Los Angeles metropolitan area oversample into the AHS-N PUF. This practice improved the precision of national estimates, while still maintaining the ability to produce precise metropolitan area estimates for these six metropolitan areas.

For the 2011 AHS, HUD and Census made a major change in how we published PUF data. To improve the precision of national and metropolitan area estimates, we merged all AHS-N and AHS-MS cases from each of the 29 selected metropolitan areas into one "combined national sample" PUF.

For the 2013 AHS, HUD and Census went back to publishing the AHS-N and AHS-MS data as separate PUFs. That survey year, there were 25 metropolitan area oversamples. To maintain consistency with the 2003 and 2009 PUFs, the Big 5 metropolitan areas were integrated into the AHS-N PUF. The other 20 metropolitan area oversamples were published in the 2013 AHS-MS PUF.

We had two main reasons for reverting to the practice of publishing separate AHS-N and AHS-MS PUFs for 2013. First, we discovered that combining the 2011 AHS-N and AHS-MS PUFs into one file required a substantial number of modifications to the geographic indicators of individual records to ensure the records conformed to the Census Bureau's disclosure rules. This process proved to be very time-consuming and potentially error-prone.

The second, and most important, reason we reverted to the practice of publishing separate AHS-N and AHS-MS PUFs for 2013 was user confusion. As many of you are aware, the AHS is a complicated survey. The combination of the AHS-N and AHS-MS samples into one PUF made the 2011 AHS itself, and the documentation necessary to describe the changes, significantly more complex.

To be frank, our decision to combine the 2011 AHS-N and AHS-MS samples into one PUF was the wrong decision. The slight improvement gained in the precision of the national and metropolitan area estimates was not worth the decrease in usability.

We are relaunching the 2011 AHS-N and AHS-MS PUFs

After a lot of reflection, deliberation, and planning, HUD and Census are re-launching the 2011 AHS PUF as separate AHS-N and AHS-MS PUFs. We're calling these new PUFs Version 2.0. While the 2011 AHS had 29 metropolitan area oversamples, to maintain consistency with the 2007 AHS-N PUF, in Version 2.0, the Los Angeles metropolitan area oversample remains integrated into the AHS-N PUF. The other 28 metropolitan area oversamples are in the AHS-MS PUF.

Practical implications of new 2011 AHS PUFs

The separation of the 2011 AHS PUF into AHS-N and AHS-MS PUFs required re-weighting every national sample and metropolitan area oversample case. Practically speaking, this means all estimates created from the new PUF Version 2.0 will be slightly different than estimates created from earlier versions. However, the changes will be small. The overall total number of housing units for the nation and for each metropolitan area that was oversampled remain the same.

Another implication of the new PUF Version 2.0 is that national sample standard errors will increase. This is because the metropolitan area oversample cases can no longer be used to produce national estimates. Again, the changes to the standard errors will be small.

No change to the 2011 Table Creator estimates

It is important to note that there will be no change to the 2011 AHS Table Creator estimates at this time. These estimates will continue to be based on a combined national sample and metropolitan area oversamples. HUD and Census are still determining the best course of action for 2011 Table Creator estimates.

Major updates to supporting documentation

The relaunch of the 2011 AHS PUFs (AHS-N and AHS-MS) requires changes to several documents associated with the survey and the use of PUFs, including:

  • 2011 Version Control Documents

  • 2011 Value Labels Package

  • 2011 Source and Accuracy

  • Geography in the PUF: 1985 – 2013

  • Metropolitan Area Oversamples History: 1973 – 2013

Look for more important announcements on AHS updates over the next few days. It's part of our ongoing effort to make it as easy as possible to access and use AHS data. As always, we want to hear from you. If you have questions or suggestions, call us at 1-800-245-2691 or email us at helpdesk@huduser.gov.

Footer

eList: