1997 AHS Data

The 1997 National American Housing Survey (AHS) represents a major departure from past surveys:

  • It is the first time that the survey was conducted using Computer Assisted Personal Interviewing (CAPI).

  • The Census Bureau changed computer platforms from a UNISYS mainframe to a UNIX workstation.

  • The Census Bureau changed its processing software/language to SAS.

This means that all the software for processing the data, editing the data, allocating missing values, recoding and transforming variables, preparing tables, and estimating variances had to be rewritten. In addition, responses are likely to be affected because of the reorganization of the questionnaire, the pacing of the interview and other modal affects.

We were very concerned about the quality of the 1997 National AHS. In order for us to do a reasonable job of quality review, we brought in outside resources (ICF, Inc. through a task order.) The task involves several sub tasks, but the major thrust of the work is to create a comparison program which operates on two columns of selected publication tables. All Households and Black Households are the two columns. The National Publication tables are 2.1, 2.2, 2.3, 2.4, 2.5, 2.6, 2.7, 2.8, 2.9, 2.10, 2.11, 2.12, 2.13, 3.14, 3.15. Supplement tables are 1.4 and 1.5. These tables cover virtually all variables and form a good cross section of data for quality control checking. The comparison tables contain about 3,000 AHS estimates.

The program compares 1997 estimates against 1995 and 1993 estimates and flags table rows where the 1997 value is;

  • less than 80 percent of the lowest of the 93 and 95 values; or
  • greater than 125 percent of the highest of the 93 and 95 values.
Once a set of comparisons were run against a file, HUD, ICF and Census met to discuss the flagged lines and to determine if there were errors that can be corrected. The corrections were then done and a new file was prepared and a new set of comparisons was run. This iterative process has gone through many cycles and has been quite successful in flagging errors and finding corrections over the last several months. Many errors were detected and corrected through this process. Some problems could not be corrected, but as a rule the variables involved were not key housing variables.

If you download the file, you should subscribe to the AHS eList. This is important since we will use the list serve to notify you when new versions of the 1997 AHS are released.

We are making available numerous items in addition to the basic data files: the comparison report, SAS programs which can be used to generate summary tables, a file of descriptive statistics, a link to the Census report, SAS code for combining the nine SAS data sets into a single household-level file, and SAS code for adding variable and category (value) labels.

American Housing Survey