Extracted from the U.S. Bureau of the Census web-site on the American Community Survey
The 1997 American Community Survey, which the Census Bureau conducted between November 1996 and December 1997, was a continuation and expansion of the 1996 ACS during the program's demonstration period.
The scope of the 1997 ACS was primarily limited to housing units, occupied and vacant, in eight sites. (Note that we conducted a procedural test of the ACS in Group Quarters in the Franklin County, Ohio site.) The eight sites, listed below, included the four sites where we conducted the 1996 ACS.
1997 ACS Sites Rockland County, New York* Brevard County, Florida* Fulton County, Pennsylvania* Multnomah County and the city of Portland, Oregon* Douglas County, Nebraska Franklin County, Ohio Houston, Texas (Harris and Fort Bend Counties) Otero County, New Mexico * Also included in the 1996 ACS
The 1997 ACS utilized the same three modes of data collection as the 1996 ACS: 1) Self-response using a mail out/mail back process; 2) Computer Assisted Telephone Interviewing (CATI); and 3) Computer Assisted Personal Interviewing (CAPI). At the beginning of each month, we mailed questionnaires to a new sample of households. The data collection for each sample was completed within a three-month cycle. A brief overview of each of these data collection modes is presented below.
1) Self-Response
The activities during the first month of the cycle involved the mailing of four mailing pieces: a prenotice letter; an initial ACS questionnaire, a reminder card, and a replacement questionnaire. We mailed the replacement questionnaire approximately three weeks after the initial questionnaire, to housing units that had not returned a completed initial questionnaire.
2) Computer Assisted Telephone Interviewing (CATI)
During the second month of the cycle, we attempted to conduct interviews via CATI with residents of sample housing units for which we had not received a questionnaire. We started the CATI phase approximately six weeks after we mailed the initial questionnaire. We conducted the CATI operations at the Census Bureau's Jeffersonville, Indiana Telephone Center.
3) Computer Assisted Personal Interviewing (CAPI) Mode
During the final month of the cycle, we used CAPI to try to obtain interviews with residents of housing units for which we had neither received a questionnaire nor conducted a CATI interview. We selected a one in three sample of these non-response housing units, and Census Bureau Field Representatives made personal visits to these sample units. If a unit was vacant at the time of interview, Field Representatives attempted to obtain basic information about the housing unit from a knowledgeable source.
We selected a systematic sample of addresses each month from the Master Address File (MAF) for seven of the eight survey sites--all except Fulton County, PA. This MAF was produced by an on-going computer match of the U.S. Postal Service (USPS) Delivery Sequence File (DSF) and the Census Bureau's 1990 Census Address Control File (ACF).
Fulton County required a different approach to data collection. We created a "pseudo" MAF through a listing-by-observation operation. From that pseudo MAF, we selected the monthly samples of addresses. In addition, the lack of a county-wide structure number/street name address system made the acquisition of telephone numbers more difficult. Because of these factors, Field Representatives delivered the survey questionnaires to the sample housing units, instead of mailing them, and we did not attempt to conduct CATI for sample households in Fulton County. Field Representatives attempted to conduct CAPI interviews with a subsample of households for which we had not received a questionnaire.
The 1997 American Community Survey helped us to refine the procedures for collecting and processing survey data. We are continuing to evaluate and modify the survey process as we prepare for the beginning of the comparison phase of the program in late 1998.
One of the most important goals of the 1997 ACS was to demonstrate the richness and utility of the data through publications, tabulations and public use files, including data on year-to-year changes for the sites which were also in the 1996 ACS. In July 1998, the Census Bureau began the release of data products from the survey sites for calendar year 1997.